Hypothalamic Substrates of Reward

Hypothalamic Substrates of Reward1
JAMES OLDS
Department of Psychology, The University of Michigan, Ann Arbor, Michigan
CONTENTS
Appetitive Behavior Produced by Electric Stimulation .......................... .556
Anatomical Locus ......................................................... .556
Methods of Measurement. .................................................. .559
Place-to-Place Behavior .................................................. .559
Operant Measures, “Self-Stimulation” ..................................... .560
Secondary Reinforcement ................................................ .56 I
Comparison Techniques .................................................. .56 I
Satiation Tests. ......................................................... .562
Qualitative Observations ................................................. .563
Parameters of Stimulation ....................................... . .......... -564
Basic Drives .............................................................. ,566
Reinforcement Controlled by Drives ....................................... ,566
Elicited Drive Responses ................................................. .566
Punishment and Reward ................................................... .567
Escape Reactions ........................................................ .568
Ambivalent Responses ................................................... .568
Anatomical Relations of Punishment and Reward Systems .................... .57 I
Interaction of Motive Systems. ............................................ -572
Physiological Interactions. .................................................. .573
Effects of CNS Damage .................................................. .573
Electrophysiological Ramifications of the Stimulation ......................... .575
Operant Conditioning of Unit Responses ................................... .576
Autonomic Responses .................................................... .57g
Psychological Interactions .................................................. .580
Arousal ................................................................ .580
Arrest ................................................................. .581
Perception ............................................................. .582
Learning. .............................................................. .582
Social Interaction ....................................................... .584
Pharmacological Interactions ............................................... .584
Effects of Drugs on Electric Stimulation .................................... .584
Direct Chemical Stimulation .............................................. .588
Speculations. ............................................................. ,592
Appetitive Behavior and the Olfactory Forebrain ............................ .5g2
Appetitive Reactions and Drives. .......................................... .592
Mechanisms of Reward .................................................. .5g4
Summary.................................................................59 5
Conclusions .............................................................. .597
Glossary ................................................................. .5g8
1 Preparation of this review and the research carried out in the author’s laboratory were
aided by grants from the Foundations Fund for Research in Psychiatry, the Ford Foundation,
the National Institute of Mental Health, the National Science Foundation, and the Wallace
Laboratories.
554
Downloaded from www.physiology.org/journal/physrev at Karolinska Institutet University Library (130.237.122.245) on February 12, 2019.
October rg 62 HYPOTHALAMIC SUBSTRATES OF REWARD 555
SYCHOLOGISTS have long identified “positive reinforcements” (i.e., rewards)

by their effects on behavior. Application of the rewarding stimulus causes
the animal, in simplest terms, to come back for more! In more formal terms
it makes the animal repeat the behavior which immediately preceded the stimulus.
Following Skinner’s (156) terminology, behaviors which can be controlled in this
way are called operant behaviors; they are distinguished from reflex behaviors
whose incidence cannot be modified by manipulation of consequences.
“Negative reinforcements” (i.e., punishments) are similarly identified by
their effects on behavior. In this case, there are both operant and reflex conse-
quences. On the operant side, there is (a) failure to repeat the punished response,
and (b) repetition of responses which terminate the punishment. On the reflex
side, there is (c) general activation of the animal during application of the punish-
ing stimulus, and (d) some set of aversive or withdrawal reflexes.
Thus, in relation to operant behavior, the application of a rewarding stimulus
and the termination of a punishing stimulus have similar effects; both cause
repetition of the preceding operant responses. This similarity has led many psychol-
ogists to postulate a single hypothetical mechanism for appetitive and aversive
behaviors (96).
In experiments used to demonstrate positive reinforcement the animal, usually
deprived of food or water or sexual stimulation, is placed in a situation where a
certain response previously decided on by the experimenter will lead to the stimulus
which has been withheld. The response initially exists in the animal’s repertory,
but it occurs rarely. The experiment is begun by placing the animal in the ap-
paratus and waiting until the response occurs at first “by accident.” Immediately
thereafter, the previously withheld stimulus is presented either mechanically or
by the experimenter. As a consequence of this correlation between response and
stimulus, either at once or after several repetitions, the response becomes more
frequent.
In experiments used to demonstrate the similar effects of negative reinforce-
ment, the animal is not deprived; instead, a noxious stimulus is presented re-
peatedly or in a steady train. As in the positive-reinforcement experiment, a
response rare at the outset is selected by the experimenter as the response whose
frequency is to be augmented. When the response occurs, the experimenter or the
apparatus terminates the train of noxious stimulation for some period of time.
After one or several repetitions of this correlation between response and cessation
of noxious stimulation, an obvious change in response frequency occurs, the re-
sponse becoming more frequent while the animal is subjected to the noxious
stimulation.
A lore and a language of experimental psychology have grown up around
this set of experimental circumstances. Of the lore, I have little to say. It has to
do with formulations of the sort which suggest that the animal “needing” the
withheld stimulus “learns” the cccorrect” response to satisfy his needs. While
much truth, no doubt, is encapsulated in formulations of this type, it cannot be
denied that such assertions often involve difficulties. There is first the problem in
separating descriptive from explanatory aspects in such assertions. There is perhaps
even greater difficulty in distinguishing concepts based on evolutionary theories
556 JAMES OLDS Volume 42
from those based on physiological research; i.e., the behavior is often “explained”
by reference to the fact that it leads to survival instead of being explained in
terms of the actual physiological mechanisms involved.
As for the language, the terms used in the present review are defined in the
GLOSSARY (see p. 598). This is meant only to identify unfamiliar usages. From
logical or linguistic points of view it is not completely accurate, but it is not en-
tirely peculiar to the present manuscript, for the intention is to identify usages on
the basis of meanings already familiar to the reader. Thus, when two words are
said to be synonymous, it is understood that they are so only roughly.
The experimental arrangements described above provide much of the subject
matter of the present review. Specifically, this review is about “operant” as op-
posed to “reflex” behavior.
APPETITIVE BEHAVIOR PRODUCED BY ELECTRIC STIMULATION

If electrodes are permanently implanted in the medial forebrain bundle of
the lateral hypothalamus of a rat and a circuit is attached which makes a brief
train of electric stimulation occur as a consequence of some response in the be-
havior repertory of the rat, that response soon comes to predominate, occurring
eventually at the maximum possible rate for the response in question (I 34). The
electric stimulation appears then to serve as a maximal source of animal gratifica-
tion; in the parlance of the behavioral psychologist, it is a source of positive rein-
forcement. If electrodes are implanted in other parts of hypothalamus or in other
structures of the brain’s olfactory system, electric stimulation often has similar but
less intense effects (I 08, I 26, I 38). In the following pages I will review the ma-
terial which has accumulated from research directed toward an understanding
of these phenomena.
The organization of the review is based mainly on a formal breakdown of the
problem into its major variables: (i) the anatomical locus, (ii) the dependent
variables, i.e., the methods for measurement used to assess the behavioral effect,
and (z’ii) the independent variables, e.g., the parameters of stimulation. The
independent variables, however, like the phenomenon of reward by brain stimula-
tion, often turn out to be major, dynamic conditions of the whole organism with
both stimulus and response characteristics. In such cases it is convenient to discuss
in the same section the effects of these variables on the brain stimulus reward
phenomenon and the reciprocal effect of the reward phenomenon on them. Such
reciprocal effects are discussed in specifically oriented sections on (iv) basic drives,
(v) punishments, and in generally oriented sections on other (vi) physiological,
(vii) psychological, and (viii) pharmacological variables.
ANATOMICAL Locus
In the rat, maximal effects are yielded by stimulation of a telencephalic-

diencephalic-mesencephalic system which forms a U in the horizontal plane. The
sides of the U are the medial forebrain bundle in telencephalon and diencephalon,
and the area under the medial lemniscus in posterior diencephalon and mesen-
cephalon, and its arch is formed by the supramammillary area and similarly
October 1962 HYPOTHALAMIC SUBSTRATES OF REWARD 557
places tissue above the whole region from mammillary body back to interpeduncu-
lar nucleus. It is not yet definitely established whether interpeduncular nucleus
itself forms a part of this system; it is quite certain that the medial mammillary
nucleus does not. Regions behind the interpeduncular nucleus have not been
systematically investigated. The structures of the mesial and medial hypothalamus
contained within the U yield, more often than not, attenuated positive-reinforce-
ment effects, but also yield negative reinforcement. Attenuated positive reinforce-
ment is also produced by stimulation in some parts of septal area, caudate nucleus,
diagonal band, some central and medial areas of amygdala, pyriform and subcal-
losal cortex (I 38).
Very mild positive reinforcement is produced by stimulation of some parts
of the cingulate cortex. Portions of the hippocampal formation (most likely the
dentate gyrus) and portions of the diffuse system of the thalamus, particularly the
anterior group, yield a questionable form of positive reinforcement that may
have on the one hand more to do with seizures than with gratification and on the
other hand more to do with secondary (learned) reinforcement than with primary
reinforcement ( I 08, I 34, I 38).
These anatomical findings are schematized in Fig. I, which presents five
coronal sections from rat brain; the most rostral, in u/$er left, passes just anterior
to the diencephalon; the most caudal, in lower right, passes just posterior to the
diencephalon.
In other mammals, the picture is similar. Brady (I 2, I 3, I 8), Nielson (105),
Sidman (155), Brown and Cohen (28), and Roberts (I 46) have studied cats.
Electrodes in lateral and half-lateral hypothalamus yield positive reinforcement
with great regularity. The strongest effects are achieved in the area of the medial
forebrain bundle. Some parts of the caudate also yield positive reinforcement.
Some parts of the septal area do not.
Bursten and Delgado (3 I ), Brady (I 8), Lilly (73-80), Brodie and his group
(25, IS), and Porter and colleagues (142) have studied monkeys. Electrodes in
medial forebrain bundle regions yield positive reinforcement of great intensity.
Paleocortical structures, e.g., orbit0 frontal and entorhinal, amygdala, caudate,
globus pallidus, lateral septal nucleus, anterior commissure, and nonspecific
thalamus are also reported to yield positive reinforcement in varying degrees on
electric stimulation.
Lilly (82) gives a very preliminary report of replication on the bottlenose
dolphin. Some electrodes in the very large brain of this animal certainly do yield
positive reinforcement, but it is impossible to determine from the report precisely
where they were placed.
Higgins et al. (58), Delgado and Hamlin (37), Heath (53), and Sem-Jacobsen
(150) have reported on humans who have had chronic electrodes implanted in
brain for therapeutic purposes. Electrodes believed to be in hypothalamus and
tegmentum have produced extreme euphoria, electrodes in septal area have in-
hibited pain and produced feelings of “well being.” Electrodes in paleocortical
regions of frontal lobe have produced milder positive reactions. Reports of experi-
ence have not been extensive: several patients with paleocortical electrodes in
frontal areas have expressed desires to marry the investigator.
JAMES OLDS Volume 42
FIG. I. Anatomical areas yielding negative reinforcement (dotted), positive reinforcement

(vertical hatching), and ambivalent effects (cross hatching) when electric stimulation is applied via
chronically implanted electrodes. Areas known to yield motivational effects but sparsely ex-
plored are also indicated (hori<ontal hatching). Positive reinforcement is definitely obtained from
many points within these sparsely explored areas. The extent of negative reinforcement obtain-
able from the same regions is less well known. The brain drawings and nomenclature follow the
atlas of Leo C. Massopust, Jr., “Diencephalon of the Rat,” in D. E. Sheer (editor), Electrical
Stimulation of the Brain, Austin : University of Texas Press, 1961, pp. I 87 fI+. Abbreviations are as
follows : ab: basilar nucleus of the amygdala; ac: central nucleus of the amygdala; ad: anterior
nuclear group of the thalamus; al: lateral nucleus of the amygdala; am; medial nucleus of the
amygdala; arc: arcuate nucleus of the hypothalamus; bi; brachium of the inferior coIIiculus; bs:
brachium of the superior colliculus; caa: anterior commissure, anterior division; cat: anterior
commissure, temporal division ; cau ; caudate nucleus; cc: corpus &llosum; ce: external capsule;
cf: column of the fornix; cJ20: optic chiasm; tin: internal capsule; da: claustrum; cn: cingulum;
co: cortical nucleus of the .amygdala; csw superior colliculus; db: diagonal band of Broca; dm:
dorsomedial hypothalamic nucleus; ds: supramammillary decussation; fbl: medial forebrain
bundle, lateral division; fbm: medial forebrain bundle; fez endorhinal groove; ff: fields of Forel;
fim; fimbria of the fornix; Jm: medial longitudinal fasciculus; fm: habenulo-interpeduncular
tract; fos: superior fornix; frh: rhinal fissure; gm: medial geniculate body; gp: globus pallidus;
h: hippocampus; hl: lateral habenular nucleus; hm; medial habenular nucleus; 2: lateral nuclear
group of the thalamus; Zat: lateral hypothalamic nucleus; Zm: medial lemniscus; m: medial nu-
clear group of the thalamus; mi: massa intermedia; ml; lateral mammillary nucleus; mm: medial
mammillary nucleus; nac. nucleus accumbens of the septum; nbo: supraoptic nucleus; tu;: neo-
cortex; ncp: nucleus of the posterior commissure; nit: interstitial nucleus; nk reticular nucleus of
METHODS OF MEASUREMENT
A first step toward demonstrating that the neural tissuesstimulated in these

experiments actually form some part of a neural substrate of ordinary positive-
reinforcement phenomena and approach behavior is to compare in a variety of
circumstances the effects of electric stimulation reward with those of ordinary
rewards.
At the outset it is well to indicate two important qualifications of all the fol-
lowing material: one is the locus of the stimulating electrode and the other is the
intensity of the stimulus. There is no single answer to any question on the effect of
electric stimulus reward because the phenomenon is achieved over a very exten-
sive set of areas in the brain with a wide range of stimulus parameters. There is
a different brain stimulus reward for each locus and for each stimulus intensity. It
is quite clear, for example, that a rat with stimulation in subcallosal cortex as
reward is quieted during the stimulation but difficult and touchy at the off (I 09);
on the other hand, an animal with stimulus in the posterior lateral hypothalamus
as reward is highly active and investigative during stimulation, “looking around
the environment with quick darting glances,” but quiet and gentle at the off (146).
Equally important differences can be made by altering stimulus intensity: a rat
pressing a bar to stimulate the hypothalamus is totally absorbed and cannot be
distracted, and therefore appears compulsive when stimulus intensities of two-
or three-times threshold are employed. At threshold intensities, however, the same
animal respondswell but now is fully alert to all aspectsof the environment and
easily distracted.
Place-to-Place Behavior
The positive reinforcement produced by electric stimulation has been demon-
strated in many ways. In the simplest experiments, the stimulus was applied
whenever the animal walked into a particular subdivision of the test chamber;
animals returned on a greater than chance basis to the place where the stimulus
was applied (3 I, 106, 107). In other experiments, a start compartment was con-
nected with a goal box by a runway; hungry rats traversed the runway faster
for the electric stimulus reward than for food (I 09). In one case a maze was
then used to connect start and goal boxes; hungry rats which received electric
stimulation as reward showed trial to trial improvement, gradually eliminating
errors and running faster. Day-to-day improvement was also demonstrated even
on the first trial of each test day, indicating that prestimulus was unnecessary to
engage the appetitive behavior (109). In the reported maze experiment, animals
the thalamus; npt: nucleus of the transverse peduncular tract; ofi: optic nerve;pc: precommissural
septal fibers; pnz: mammillary peduncle; Pp: cerebral peduncle; pu; putamen; pyc: pyriform cor-
tex; s: septal region; sb; bed of the terminal stria; sm; medullary stria of the thalamus; sma; rostro-
ventral fibers of the medullary stria; SK substantia nigra; SOC: supraoptic commissures; sfl: com-
missural bundle of the terminal stria ; to: olfactory tubercle; tol: lateral olfactory tract; tpt: trans-
verse peduncular tract; tro: optic tract; v: ventral nuclear group of the thalamus; vda: mammil-
lothalamic tract; vk lateral ventricle; vm: ventromedial hypothalamic nucleus; si: zona incerta.
560 JAMES OLDS Vohim 42
learned somewhat faster for food reward but ran faster for the electric stimulus.
Electrodes were in a telencephalic area; later work shows that with some dien-
cephalic electrodes animals learn even faster (I 34).
Operant Measures, “Self-Stimulation”

By far the most widely used method of measuring positive reinforcement is
Skinner’s operant technique (I 56). A circuit is arranged to make the reinforcing
event occur as a regular consequence of some simple response in the animal’s
behavior repertory (I 08, I 26). The response in this experiment was termed “self-
stimulation” by Brady (I 4, 16), and this term has been used regularly. Response
rate is measured during an “acquisition” period when the reinforcing event is
applied; or during an “extinction” period when, after a specified acquisition
period, the reinforcing event is withheld. Both acquisition and extinction
rates when compared with “operant” rates provide measures of the intensity of
the positive reinforcement. The operant rate for a given animal in a given appara-
tus is the random response rate before acquisition, or after an extensive period of
extinction. When a lever-pressing response is used with rats, many factors, in-
cluding the size of the testing chamber, the size of the lever, and the effort of the
response, determine the operant rate.
In a box with operant rate below 50 responses per hour (rph), acquisition
scores often ranged higher than 8,000 rph when medial forebrain bundle stimula-
tion was used as reinforcement (I 34). In similar experiments with monkeys,
rates of I 7 responses per second have been reported for brief intervals (74). In
rats, rates above 5,000 rph have regularly appeared with electrodes in medial
forebrain bundle structures of lateral hypothalamus; rates of 300 to I ,000 are the
rule in telencephalic areas.
In the first reported extinction tests, responding often dropped off very rapidly
after withdrawal of reinforcement, so rapidly that this was thought to indicate a
difference between electrical reinforcement and reinforcement with food or other
more ordinary primary reward (I 26, I 51, I 55). This, however, turned out to be a
function of locus of stimulating electrodes (25). With telencephalic electrodes,
extinction did occur more rapidly than would be expected considering the very
high rates of responding during acquisition (I 26, I 51). By the same token, animals
with telencephalic electrodes did not sustain very high ccresponse ratios,” i.e.,
they would not respond if required to make 50 lever responses for just one reward
(I 55). Very high ratios are more easily maintained by food rewards (42).
It is now shown (25) that whatever the reason for the rapid extinction and
low ratios observed with telencephalic stimulation as reinforcement, the same does
not hold true with diencephalic stimulation in medial forebrain bundle used as
reward. A very large response output in extinction, and ratios higher than IOO
responses for just one reward, are reported. Another study (69) indicates that with
stimulation in posterior hypothalamus as reinforcement, extinction is slower with
electric stimulation as reward than with food as reward.
With electrodes in the very powerfully reinforcing medial forebrain bundle
area, a disrupting effect has been observed on the timing behavior involved in the
Ferster and Skinner (42) DRL (differential reinforcement of low rates) program
(15, I 7, 20). In one case the program was arranged so that only the first response
after each ao-second pause would be reinforced. A monkey working for sugar
pellets eventually spaced responses allowing quite often the required 20 seconds to
elapse between responses, and thus obtaining a reward quite often. The same
monkey working on alternate days for medial forebrain bundle stimulation ap-
peared unable to pause; it responded with inefficient rapidity giving about IOO
responses for each reinforcement and a wasted burst of responding after each
reward which lasted for about IO minutes.
Secondary Reinforcement
Several other dependent variables have been used, first with the purpose of
showing that the brain stimulus reinforcement has properties of other well-known
primary reinforcers and second with the aim of establishing the intensity of the
motivation involved and comparing intensities of stimulation at different brain
points.
Among these is the “secondary reinforcement” experiment. A food or sexual
reward not only has the power to motivate behavior but also it can impart
motivating power to formerly neutral stimuli with which it becomes associated.
The dog comes to a whistle because this “secondary reinforcement” has been
associated with primary reward. Stein (I 57) showed that a neutral tone associated
with a brain reward stimulus would in a similar fashion acquire reinforcing value
for the rat; and he speculated on the possibility that the tone-stimulus pairing
gave the tone some power to elicit activity in the neural tissue near the electrode
tip.
Comparison Techniques
The obstruction box experiment is normally used to compare the intensity of
different positive reinforcers by matching them against a measurable negative
reinforcement. The animal is required to cross a grid which yields a quantifiable
foot shock in order to get positive reinforcement. In a box where hungry rats,
starved for 24 hours, would take a 60- to 18o-pa foot shock for food reward,
an implanted animal took 60 pa of foot shock for stimulus of twice the threshold
value in medial forebrain bundle, and a 4250pa foot shock for a stimulus of ten-
times threshold in medial forebrain bundle (I I 5, I 3 I ).
The obstruction box also provides one method for comparing the different
intensities of reinforcement produced by stimulation at different places. It was
thought that even though a rat responds more slowly for telencephalic stimulation
than for medial forebrain bundle stimulation, the former might be more rein-
forcing; e.g., it might have a long enduring effect which would slow the
animal down. In one experiment a group of animals with telencephalic electrodes
was compared with a group with electrodes in diencephalon. The same brain
stimulus intensity was used in both cases. Within each group intensity of rein-
forcement as measured by grid crossing was directly related to intensity as meas-
ured by response rate. The hypothalamic stimuli which showed greater intensity
by producing higher response rates also showed greater intensity by causing rats
to traverse the greater shock obstruction. However, the most intense telen-
cephalic reward placements produced lower response rates but better grid crossing
than the least intense hypothalamic placements. Thus it is evident that response
rates, while useful, are not perfect for measuring positive reinforcement (I 14,
131)*
A similar point is made by a series of preference experiments (I 8, 60, I 59). In
the first of these, advantage was taken of the fact that a monkey may balk or slow
down if a lesser reward is offered after a more intense one (I 67). Testing the same
animal with the rewarding brain stimulus delivered first to one area, then to
another, Brady (I 8) varied the order of the points tested from day to day. He found
that the animal would not work for amygdaloid stimulation after medial forebrain
bundle stimulation, anterior medial forebrain bundle after hypothalamic stimula-
tion, and so forth. With this technique he worked out a hierarchy of structures
according to the reinforcing value of stimulation with the following order from
maximum positive to negative reinforcement : hypothalamus, anterior medial
forebrain bundle, orbitofrontal cortex, amygdala, entorhinal cortex, septal area
(neutral), and fornix (negative).
A more recent work (60) criticizes rate as a method of measuring the rein-
forcing value of stimulation at different brain points. It shows that the brain
stimulus producing highest rate is not always the one preferred in a choice test,
that is, if mild stimulus in a preferred area is compared with strong stimulus in a
nonpreferred area. However, even here, the brain locus yielding the greatest
reinforcement by rate test is the same as that producing the greatest reinforcement
by preference test. The most interesting outcome of the study has to do with high-
intensity stimulation. As intensity of stimulation increases, rate sometimes declines
(I 34, I 44), as will be discussed later. The preference study shows that even when
rate declines at higher intensities, choice is still sometimes directed to the higher
intensity.
Applying a similar technique, Stein (I 59) permitted concurrent stimulation.
Animals with two pedals could work them alternately or in any order they might
choose to stimulate two different brain points. He showed that rats maintaining
high rates on each electrode separately could be made to work twice as hard to
stimulate the two concurrently. For example, when presented with lever A
alone to stimulate ehtrode A, one rat regularly maintained a 1,5oo-rph rate.
When offered at the same time lever B to stimulate electrode B, the rat maintained
the 1,500 rph on A and by racing back and forth and working twice as hard main-
tained a similar rate on lever B concurrently. Behavior at one pedal was often
largely independent of behavior at the other, although interesting interactions
were shown in certain cases. Posterior points in hypothalamus usually required
less current for the same rate than that required by anterior ones.
Satiation Tests
Another test has had to do with endurance. Sometimes behavior motivated
by brain stimulation has been sustained for periods of more than 24 hours at a
time. It has been shown that, in long-run self-stimulation tests (I I 3), animals
with diencephalic electrodes tend to respond to the point of physical exhaustion.
Animals with telencephalic electrodes tend to become satiated long before the
state of physical fatigue brings self-stimulation to a halt. In both cases, however,
there is no satiation if animals are permitted only I hour of self-stimulation daily.
Qualitative 0 bservations
Several investigators have discovered elicited effects taken to be indicative of
positive effects. MacLean (85) reports that stimulation of hippocampus by deposit
of crystalline carbachol induces seizures in the cat which subside after an hour or
so. During subsidence “enhanced pleasure and grooming reactions” are observed
and the cat is “unusually receptive to genital stimulation.” The pleasure reactions
include ccspontaneous loud purring, kneading and fanning the forepaws; turning
over on the back, rolling, twisting, stretching . . . ; rubbing head and body vig-
orously against inanimate objects or the examiner.”
Kopa, Szabo, and Grastyan (71) report “a general relaxing effect” from stimu-
lation in centrum medianum of thalamus, and also sometimes activation of “ali-
mentary reflexes.” Further material on activation of instrumental and consumma-
tory responses in conjunction with positive reinforcement will be taken up
later in the section on basic drives.
The small number of studies with humans have made reference to the pa-
tients’ report of experience. Possibly because of the unusualness of the experiences
involved, these reports have not been unambiguous. Sem-Jacobsen (150) reports
“feeling of ease and relaxation, feeling of joy with smiling, and great satisfaction.”
Further, he speaks of “desire for repeated stimulation” and experiences “ranging
from curiosity and funny tickling to relaxation and pleasure.” He tells of one pa-
tient in whom “the stimulus evoked a fluttering in a muscle group in the pelvis
which tickled the patient and she responded with joy and laughter.” He allowed
patients to stimulate themselves by pressing a button and found that Yn some
regions they like to keep the stimulus on for a prolonged period, only interrupted
by short breaks. In other areas, they seem to get pleasure by frequently starting and
stopping the stimulus.
“The most rapid rate of pressing and releasing the button was obtained when
the patients’ level of consciousness was altered in connection with self-stimulation.
Frequently, as long as they were unresponsive and after discharges appeared in
the record, they would press and release the button with a high repetition rate.
Afterwards they were unable to explain the behavior. We have never obtained
anv results similar to the rapid rate of IO per second or more into which animals
stimulate themselves.” It should be pointed out that in fact animals rarely exceed
2 or 3 responses per second, and typical rates with telencephalic electrodes are of
the order of one response every several seconds.
Finally, Sem-Jacobsen indicates that “from strong pleasure areas we have
found that the patients stimulate themselves into a convulsion. In the postictal
stage these patients were lying relaxed, smiling happily, contrary to the restless
fighting frequently observed in patients after electrotonic treatment.”
PARAMETERSOF STIMULATION
It is assumed that the negative phase of the electrical stimulus is the effective
phase. This can be square, peaked, or rounded. It can be brief or extended in
time. It can be single or repeated. If repeated, as it usually is, the repetition rate
can be varied over a very wide spectrum, the train duration becomes an important
variable, and the repetition rate of trains can also be important.
Most important, of course, is the factor of intensity, but the measure of inten-
sity poses problems. Is it peak current that counts, or the total amount of electricity
in a pulse, or in a train. 3 If the total electricity in a train is important, how long
is the interval of summation?
In the experiments reported, trains of repeated pulses have been used regu-
larly although it is established that a rat will work with extreme rapidity
for single pulses of several milliseconds duration (Tullsen and Olds, unpub-
lished). Thr ee general wave forms have been used regularly and with considerable
consistency between investigators; these are rectangular pulses, “Lilly” pulses(8 I ),
and sine waves.
Rectangular pulses ranging from o. I to 2.0 msec in duration have been reported
with repetition rates of 50 to 300 pulses per second (pps) and intensities of about
0.5- to 3.o-ma peak (25, 26, 31, 38, 60, 105, 146). Paired positive and negative
triangular pulses of the “Lilly” (81) type with negative pulse ranging from 0.025
to 0.050 msec in duration have been reported with repetition rates of 50 to IOO
pps and peak intensities of about 12 to 50 ma for this very brief, saw-toothed-
like pulse (I 8, 74, I 55, I 57). Sine waves with negative pulses ranging from 2 to 8
msec, these determined inversely by repetition rates ranging from 60 to 200 pps,
have been reported with intensities of about 0.005 to o. IO ma r.m.s., which yields
about 0.007 to o. I 4 ma as the peak for the negative phase (I I, I 26, I 44). Brown and
Cohen (28) reported use of a 5,ooo-cycle/set sine wave with current of I .3 to 2.1
ma.
In a study of parameters, Ward (I 70) used 0.2.set trains of rectangular stimuli
with frequencies in the ao- to IOO-pps range and pulse durations from o. I to 5.0
msec. Peak current appeared to be the most important variable, total coulombs
next, and pulse width and frequency only insofar as they contributed to these. For
trains of constant pulse width, the number of coulombs per train determined
thresholds; frequency and intensity were both significant because their product
determined the number of coulombs in the train as a whole. When pulse widths
were varied, it was found that peak current flow was somewhat more important
than the total amount of electricity in a pulse: a 3-fold drop in peak current re-
quired a IO-fold increase in duration to compensate, causing thereby more than a
3-fold increase in the total electricity required to produce an effective stimulus.
From the point of view of saving coulombs, therefore, very brief stimuli were most
effective; more enduring stimuli, however, permitted considerable reduction in
peak intensities.
Ward (I 70) also tested sine wave trains of o.g-set duration, o.og-ma intensity,
and frequency ranging from I 7 to 5,000 cycle/set. Frequencies from 40 to 1,000
October rg62 HYPOTHALAMIC SUBSTRATES OF REWARD 565
cycle/set were effective in yielding positive reinforcement with an optimum some-

where in the 40- to 2oo-cycle/set range.
Several studies have involved manipulation of electric current with other
parameters constant. These show clear-cut thresholds, and a variety of functional
relations as current is increased beyond threshold levels (I 34, I 44, I 55). With
electrodes in some medial forebrain bundle locations, response rate is augmented
by each increase in electric current even up to I 6-times threshold (I 34). With other
electrodes in some septal locations, responding never rises above the low levels
which appear at threshold; in these cases the same rates appear at threshold and
at I 6-times threshold. With electrodes in ventromedial hypothalamus (144), re-
sponse rates are augmented by stimulus increases up to some relatively fixed cur-
rent level; further increases cause response rates to decline more or less steadily.
In these cases, the optimal current level is relatively stable for a given electrode
placement. In preference studies (60) it appears that sometimes these animals may
prefer the higher intensity, even when they respond more slowly.
Self-regulatory experiments (I 61), however, show that, with posterior hypo-
thalamic electrodes, animals often prefer less than maximal current levels, whereas
with telencephalic electrodes, animals often prefer the maximal current levels,
levels which regularly produce seizures. Even with posterior hypothalamic elec-
trodes where preference is for less than maximal stimulus intensities, the current
levels preferred are very high, much higher than an experimenter would normally
select for an experiment.
In another form of self-regulatory experiment, animals can be trained to indicate
a threshold level by their response (I 62). The electric stimulus is started at a
maximum level; each response causes a brief stimulus to the brain and also causes
the stimulus to be set down one step. When the current becomes too low to satisfy,
with slight additional effort the animal may press a reset lever which starts the
cycle again with current at the prefixed maximum level. Th stimulus level at the
time of reset is taken as the threshold level. It is quite stable over time for a given
animal and is a dependable function of pharmacological variables.
Preliminary material is available on train duration (Stein, private communica-
tion; I 2 I). When electrodes are in posterior hypothalamus, animals usually select
very brief trains of considerably less than I second. When electrodes are placed in
septal region, trains of I to g seconds are selected. With electrodes in cingulate
cortex, trains of indefinite duration are selected. Roberts (146) shows that, with
some hypothalamic electrodes, very long trains (3 min) are aversive, while short
trains (0.5 set) yield positive reinforcement; with other hypothalamic electrodes,
even 3-minute trains yield positive reinforcement.
As for the repetition rate of trains, it is clear that positive-reinforcing trains can
become negatively reinforcing if applied too frequently (I 20). It also appears,
however, that positive reinforcement sometimes becomes attenuated if trains are
too widely spaced: animals move in a runway faster for brain stimulus reward if
trials are spaced 20 seconds apart than if I 5 minutes intervenes (I 52).
It is obvious from this review of the very preliminary work available that a
more systematic study of stimulus parameters can be expected to contribute greatly
to understanding of these positive-reinforcement effects.
BASIC DRIVES
The relation of these phenomena to hunger, thirst, sex, or other drives is an
important adjunct to the more basic question of whether these electrodes penetrate
into real physiological substrates or elicit veritable central components of normal
reward mechanisms. Normal reward mechanismsare ordinarily responsive to the
basic drives, and if the electrically aroused brain processesare veritable substrates,
they should respond similarly.
Reinforcement Controlled by Drives
Hunger drive is definitely established as influencing the brain-rewarded re-
sponserate when electrodesare placed in the feeding center of lateral hypothalamus
of rat (61, 93). Hunger also controls the response rate when electrodes are in
medial septal area of rat (I g, 59, I I 2) and ventromedial caudate of cat (I g, 105).
Thirst may be effective also with electrodes in these anterior points (IS), but
this is not clearly established as separate from the hunger effect.
Castration and androgen replacement therapy have exercised more or less
complete control over self-stimulation rates with electrodes in the dorsomedial
caudate or the half-lateral, middle region of the reticular activating system (I 15),
and milder control with electrodes in the olfactory tubercle region of the diagonal
band and in the supramammillary regions of the hypothalamus (I I 2).
It is established that the hunger-related rates produced with electrodes in
medial septal sera of female rats are not susceptible to augmentation by the
hormonal manipulations which augment the sexual reflexes (59). Similar dissocia-
tion of hunger and androgen effects is reported for male rats (104, I I 2). In general,
it appears that placements yielding self-stimulation susceptible to positive control
by hunger drive are different from placements similarly related to the sex drive
(I 12).
Elicited Drive Responses

A different and equally important relation between the basic drives and the
reward mechanisms appears from a different series of experiments in which
electrodes are first tested for elicitation of consummatory or other drive-related
responsesand then for rewarding effects.
It has long been known that electrodes in lateral hypothalamic centers elicit
eating responses,and that lesions in these areas can cause more or lesscomplete
cessationof eating (2-4, 36, 56, I 02, I 65). Only recently with increasingly detailed
mapping has it become clear that this same lateral hypothalamic sector is one of
several regions yielding maximal positive reinforcement on electric stimulation (I 34,
I 73). It is now shown (93, IOO) that many electrodes which yield eating responses
also yield positive reinforcement and that the threshold stimulus level may be the
same for the two kinds of response.
Another study by Hoebel and Teitelbaum (61) also shows stimulus-bound
eating and hunger-related self-stimulation with electrodes in lateral hypothalamus
and that stimulation in the satiety center (cf 23, 1or> of medial hypotIx&um.ls
eliminates both the eating and the self-stimulation responsesof the lateral hypo-
thalamic center. Lesions of the medial satiety center, on the other hand, augment
both the feeding and the self-stimulation caused by lateral stimulation.
These findings appear to indicate a very near identity of the lateral mechanisms
responsible for lowering the threshold of eating reflexes and that responsible for
hunger-related positive reinforcement.
A similar relation between sexual mechanisms and self-stimulation centers
waits to be demonstrated. However, a great deal of work (27, 40, 41, 43, 86-90,
148) implicates a variety of nervous system structures in various sexual phenomena.
Included are the hippocampus, the septal area, the dorsal hypothalamus, and the
supramammillary region. All these areas are also implicated by positive reinforce-
ment tests; and at least the supramammillary region (I I 2) appears to yield andro-
gen-related positive reinforcement.
As for thirst, a region in posterior hypothalamus of goat, near the mammillary
level but placed just lateral to a line from mammillothalamic tract and fornix
yields voracious drinking on electric stimulation (7). The homologous point in
rat yields very intense positive reinforcement (I 38).
In the studies of hunger (49, 96) and thirst (6) stimulation of the rewarding
lateral hypothalamic CCdrive” center not only calls forth consummatory behavior if
the goal is presented but also specific, learned, goal-directed instrumental responses
if the goal is absent.
A generalization which is growing in likelihood is that at a brain point where
electric stimulation lowers the threshold of instrumental and consummatory re-
sponses appropriate to a given drive, there also stimulation will yield rewarding
effects whose intensity varies as a function of the same drive. It is tempting to
suppose that various drive-response systems exist; in each case, the threshold of
the system varies with appropriate hormonal or deprivational conditions. Activity
in the system lowers the thresholds of related instrumental and consummatory re-
sponses; and strong arousal of the system yields positive reinforcement of preceding
behaviors.
PUNISHMENT AND REWARD
The characteristic behaviors associated with the primitive property of irrita-

bility inherent in living matter are negative avoiding reactions. These aversive
reactions are far simpler to explain on a cause-effect basis than are appetitive or
homing reactions. Possibly because of this, a tendency to parsimony has led to
many attempts to show or suggest that the seeming appetitive reactions are nothing
but aversive reactions in disguise (96, 98, gg).
Direct experience, on the other hand, at first makes it appear that reward and
punishment are quite different mechanisms for controlling behavior. Yet even in
experience, these mechanisms rarely appear in conflict; pleasure and pain are
rarely reported simultaneously, and the same behavior often seems aimed at the
avoidance of punishment and the pursuit of reward. If the mechanisms are dual,
some method of interaction or reciprocal correlation seems to be worked out within
the organism to prevent conflicts.
Experiments in which aversive reactions are produced by electric stimulation
of the brain have yielded quite definite information about anatomical structures
peculiarly related to mechanisms of negative reinforcement. These have recently
been combined with studies of positive reinforcement to further the analysis of
relations between mechanisms of punishment and those of reward.
Many studies which show elicitation of appetitive and aversive reactions from
stimulation of the same point have been taken as evidence for a single- rather than
dual-motive mechanism; studies which show differentiation of appetitive from
aversive points have been taken as evidence for a dual-motive mechanism. Studies
of interactions suggest possible mechanisms of reciprocal correlation.
Escape Reactions
In early reports, stimulation of posterior hypothalamus and mid-line thalamic
nuclei in cat was reported by Hess (56) to elicit a pattern of attack-defense. Recent
work on cat and monkey (35, 38, 39, I 45-147) indicates painlike responses and
avoidance responses from a variety of midbrain areas, including medial lemniscus,
spinothalamic tract, central gray, trigeminal nerve and its root, and from a ventral
tectal location near the posterior commissure. Similar responses have also been
reported from related structures in thalamus, namely, ventral nucleus and possibly
also lateral and dorsomedial nuclei. A fearlike response characterized by avoidance
behavior has been definitely reported from stimulation in the dorsomedial thalamic
nucleus ( I 47). Similar responses have also resulted from electric stimulation applied
to hypothalamus near the ventral aspect of the posterior hypothalamic nucleus,
and in the Fore1 Hl field of zona incerta. Rage has been produced with electrodes
in the ventromedial nucleus of hypothalamus, in the fornix, and sometimes in
mammillothalamic tract (94, 95, I 46). In telencephalic systems, hippocampal
points, fornix system points, and parts of the amygdala have been implicated in
negative reinforcing effects of electric brain stimulation (ESB) (35, 38, 39). In the
rat (138), the area of pure negative reinforcement is best described as it appears on
one half of the transverse of coronal plane (see Fig. I). In the midbrain, it appears
to form a full circle surrounding the reticular activating system. The circle is
formed by the brachium of the medial geniculate on the lateral boundary, by the
medial lemniscus below, by the periventricular gray on themedial edge, and by the
ventral tectal structures on top. In the thalamus, the system seems to have lost its
upper boundary and thus it now forms a U. At this point, lateral forebrain bundle
and some parts of the reticular nucleus form the outer edge, the medial lemniscus
still forms the base, and the mid-line structures of thalamus form the medial border.
In telencephalon negative reinforcement appears somewhere on the boundary of
the lateral ventricle (apparently in caudate), and in the vertical fornix column and
in hippocampus proper. (Dentate gyrus does not appear to yield negative reinforce-
ment.) Stimulation of ventromedial nucleus of hypothalamus has also produced
unequivocal escape behavior (I 20).
Ambivalent Responses
Roberts first reported rewarding and punishing effects of stimulation of the
same electrode at the same intensity. He came upon the effect while investigating
the area of the lateral boundary of the posterior hypothalamic nucleus. He found
that electric stimulation caused escape behavior after the onset, but that the animal
would not heed a warning signal and avoid prior to brain stimulation (I 45).
Roberts first guessed that for some reason the brain stimulus failed to become
associated by normal learning mechanisms with the warning signal. Later (146),
however, he tested the notion that the animal might be at first rewarded, by the
onset of the stimulation, and then punished, by its continuation. Proceeding on
this assumption, he found that animals would press a lever to turn the stimulation
on, and would also respond to turn it off. Later, using a symmetrical Y maze with
one alley for “on,” one for “off,” and one for leaving the stimulus CCasis” whether
on or off, he found that these animals would work to turn on and then to turn off
the same stimulus. At low intensities the turn-on response was dependable and
the turn-off response nearly random. As the intensity increased, the turn-off
response became dependable and the turn-on response became slower and more
conflicted.
His assumption therefore was that brief or low-intensity stimulation was posi-
tively reinforcing, but with increased intensity or prolonged duration, the positive
reinforcement became less and a negative reinforcing component of the stimulus
appeared.
In this as in the other approach-escape experiments, identical or roughly
similar stimulus intensities were used in reward and punishment tests. Character-
istically, however, the train duration was fixed at some brief level during reward
experiments but was continued until response occurred in escape experiments;
thus the duration was longer in the escape or avoidance experiments.
In the same study (146) a special test was made to assess the duration factor.
Animals were forced to take a s-minute train of stimulation or none at all. Under
this regimen, two animals showing milder reward in previous tests now chose
none at all, while one animal which previously showed strong reward now took
the s-minute stimulus. For two of the animals, therefore, extending the duration
of the stimulus transformed it from positive to negative.
The one case rewarded by the longer train must be taken seriously, however,
for it makes an important point, namely, that the onset of the hypothalamic stimu-
lus was in itself a rewarding event. Some earlier arguments suggested that animals
pressing for such stimulation were in fact rewarded by its cessation.
The work of Roberts was followed by that of Bower and Miller (I I) who re-
ported that rats with anterior medial forebrain bundle electrodes would work
both to approach and to escape from electric stimulation, but that rats with elec-
trodes in the posterior part of this same bundle showed pure approach behavior.
Brown and Cohen (28) implanted electrodes in dorsal, half-lateral areas of
hypothalamus of cats at a point dorsal to the ventromedial nucleus; their points
yielded classical “hypothalamic rage.” They showed that cats would respond
faster on successive trials to get a 0.3.second train of stimulation at these points,
and also would escape from the same stimulus when it was continued up to the
time of the escape response. These animals, unlike those of Roberts, did learn to
heed a warning signal, and eventually many of them responded early enough to
get no stimulation at all. The authors conclude that the stimulus had merely an
activating effect without appetitive or aversive characteristics. But the data are
possibly better interpreted by Roberts’ (146) assumption of rewarding effects
associated with brief stimulation and aversive effects associated with more enduring
stimulation at the same points.
Analyzing the tegmentum of rats, Olds and Peretz (130) found that dorso-
medial points and medial lemniscus points caused animals to escape from brain
stimulation onto an aversive foot grid, and caused no appetitive responses. Stimu-
lation in ventrolateral tegmentum caused no escape response but strong appetitive
behavior, and stimulation in middle parts of the reticular activating system caused
both the escape and the appetitive responses depending on the nature of the test.
Utilizing a technique which permitted the same animal to press the same pedal
first to turn on electricity in hypothalamus, then to turn it off, and later to turn
on electricity in tegmentum, then to turn it off, it was shown (120) that
lateral hypothalamic electrodes in the medial forebrain bundle, and other elec-
trodes in the anterior commissure region, would yield reward but not punishment;
electrodes in ventromedial nucleus of hypothalamus and in dorsomedial tegmen-
turn yielded escape but not reward. Some placements in half-lateral hypothala-
mus yielded both. In this case, the escape stimulus was more nearly identi-
cal with the one used in the reward studies as both had the same train duration;
however, in escape studies, the trains were more frequent. The repetition of trains
occurred at a rate of I per second unless stopped or postponed for 4 seconds by a
pedal response (cf I 54, I 68). In reward tests, response rates of ambivalent rats
were never above one response for every 2 seconds. Thus it appears that having
the stimulus applied too often was aversive in this case.
Applying this dual-test technique to map the diencephalon in rat (138) (see
Fig. I), it was shown that electrodes on many of the boundaries of the positive-
reinforcement system yield attenuated positive reinforcement but also yield
escape responses. The main region, however, in which electrodes produced these
ambivalent reactions is the group of nuclear masses which make up the medial
hypothalamus; all of the medial hypothalamus is involved. It was at first thought
that only boundaries between appetitive and aversive areas would yield the
ambivalent reactions, but it is difficult to treat the whole medial area which has
long been considered the main body of the hypothalamus as a boundary region of
the medial forebrain bundle. In any event, because it is more than 2 mm
across, and because ambivalent reactions occur right in the middle, it is no
longer possible to contend that ambivalent reactions occur only on the boundaries
of the pure positive system. The ambivalent reactions also occur with considerable
frequency when electrodes are placed anteriorly in the medial forebrain bundle
itself (I I, I 38); in fact, pure positive cases are extremely rare with electrodes
in the anterior hypothalamus or anywhere in the telencephalon. This suggests
that the positive system may be more diffuse and intermingled with other
systems in these anterior areas.
In all the ambivalent cases reported, it appears that the stimulus becomes
aversive if presented immoderately. This view is in harmony with work mentioned
October rg62 HYPOTHALAMIC SUBSTRATES OF REWARD 57=
earlier showing that with electrodes in certain places a decline in positive reinforce-
ment sometimes occurs at high electric stimulus levels (60, I 44, I 6 I). We may con-
clude that with electrodes in certain places a change from appetitive to aversive
effects often occurs on the basis of changes in the quantity of stimulation, i.e.,
either changes in duration of train, number of trains per unit time, or intensity of
stimulus; there is a possibility that changes in duration are more important than
changes in intensity.
Two papers suggesting change of reinforcement sign based on external factors
have appeared. Nielson et al. (105) indicate that using a neutral caudate stimulus
as a warning signal of oncoming aversive shock converts the caudate shock to an
aversive stimulus itself. Kopa, Szabo, and Grastyan (7 I ) report that stimulation in
diffuse thalamic areas causes increased fearlike behavior in an otherwise dangerous
situation, and increased relaxation in an otherwise safe situation.
In summary, for some cases the prime determinant of reinforcing effects is
the locus of the stimulating electrode. Thus, stimulation in the anterior commis-
sure region (74, I 20) and in the medial forebrain bundle (26, I 20) seems irreversi-
bly positive in reinforcing effects. Stimulation in the periventricular fibers of
dorsomedial tegmentum (35, 39, I 20, 130, I 38), in the medial lemniscus (35, 38,
146), in ventral thalamus (38), in dorsomedial thalamus (147) and possibly in
meningeal or chiasmal regions below anterior commissure (74) is irreversibly
negative.
For other points, particularly in medial hypothalamus, the amount of stimu-
lation seems the prime determinant of reinforcing effects with brief and low-
intensity shock yielding positive reinforcement and high-intensity or long-enduring
shock becoming negative in reinforcement sign.
Finally, some points in caudate and in diffuse systems of thalamus may take
on reinforcement sign either from associative learning or from other apsects of
the situation. Other points in paleocortical, amygdaloid, and paraolfactory
regions are yet to be explored to determine the extent to which the reinforcement
sign is reversible. Work of Wurtz (I 76) suggests that there, too, some points will
be irreversibly positive, others irreversibly negative, and still others changeable on
the basis of stimulus or situational or associative factors.
Anatomical Relations of Punishment and Reward Systems

What can be concluded from the set of anatomical relations between points
yielding positive reinforcement and points yielding negative reinforcement? At a
very gross level, the most striking fact is the enormous difference between hypo-
thalamus and thalamus. There is almost no report of unequivocal escape behavior
produced by electrodes in hypothalamus; and very little report of persistent
self-stimulation with electrodes in thalamus. Even though these statements are
only approximately true, there is no doubt about the fundamental difference. At
the very least, one is led to wonder about the evolutionary and functional signifi-
cance of an arrangement which appears to put negative reinforcement mechanisms
mainly in thalamic systems and positive mechanisms mainly in hypothalamic ones.
At the level of detail, there are three surprising points (138): (i) the close
572 JAMES OLDS Vomm3 42
synaptic relation of these apparently different motive systems to one another; (ii)
the tendency to find “pure” effects in fiber bundles, and “ambivalent” effects in
nuclei; and (iii) similar thresholds and electric current functions for appetitive
and escape behaviors in mid-hypothalamic locations.
It appears quite possible that fiber bundles yielding positive reinforcement
regularly synapse with those yielding negative reinforcement. Most definite are
the findings (138) that the region of the medial forebrain bundle which is the
primary input to mid-line hypothalamic nuclei yields positive reinforcement, that
the nuclei themselves yield ambivalent effects, and that the periventricular system
of fibers, which appears to be the main outflow of medial nuclei, yields pure
negative reinforcement. Similar but inverse patterns may exist at mammillary
and habenular synapses but these remain to be definitely validated. The hypothesis
of inversion of sign from input to output of hypothalamic nuclei is strengthened
by +zts ii and iii above. Ambivalent effects in such a case would be achieved
from stimulation of the nuclei themselves because the stimulus would affect both
afferents and efferents. And the field of afferents and efferents would be relatively
homogeneous, thereby accounting for the similarity of thresholds and functions
for the two effects.
It has been proposed that a large number of direct synaptic relations between
elements whose stimulation yields effects of opposite sign might indicate that one
or several of the main projection pathways in this group of systems have inhibitory
rather than excitatory function.
From the anatomical data, one is led to think of a Papez-like circuit (139)
consisting of alternating P fibers (whose stimulation is positively reinforcing) and
N fibers (whose stimulation is negatively reinforcing), all more or less spontaneously
active and each exerting an inhibitory influence on the next neuron of the circuit.
If such a system existed, it would function to mediate reciprocal inhibition of
positive and negative reinforcing systems, and it would indicate the likelihood of
some common mechanism of action between the two systems.
To indicate the totally hypothetical nature of these present inferences, how-
ever, it is well to point out immediately that, while the data above seem to suggest
a one-way inhibition of medial forebrain bundle on the periventricular system of
fibers, data to be presented shortly seem to suggest just the opposite. The exact
relations involved remain to be explicated by further experiments.
Interaction of Motive Systems

In one set of studies (129) animals were tested for positive self-reinforcement
during a continuous negative stimulation; and for self-induced escape behavior
during a continuous positive stimulation. The test of positive self-reinforcement
was made by presenting a hypothalamic stimulus after each lever response, during
a I-minute period while a constant train of stimulation was being applied to the
negative reinforcing point in dorsomedial tegmentum via a different electrode.
The test of negative reinforcement was made by presenting repeated $$second
trains of tegmental stimulation at a rate of I per second and interrupting this
sequence for 4 seconds after each lever response, and this was done during a 2-
minute period while a constant train of stimulation was being applied to the posi-
tive reinforcing point in lateral hypothalamus via a different electrode.
The constant negative train, as might have been expected, impeded and some-
times inhibited completely the positively reinforced behavior. A similar finding is
reported by Hoebel and Teitelbaum (61) with stimulation in the ventromedial
hypothalamic “satiety” center where stimulation also causes negative reinforce-
ment (I 20). Ventromedial hypothalamic stimulation inhibits self-stimulation via
lateral hypothalamic electrodes; and also, lesions at the ventromedial point cause
an augmentation of the positively reinforced behavior (6 I).
Quite surprisingly, however, the constant positive train, far from impeding
the escape response regularly facilitated the negatively reinforced behavior (I 29).
The asymmetry of the outcome appears to indicate a one-way inhibition acting
from the escape mechanism of the periventricular area onto the reward system of
lateral hypothalamus. In an interesting fashion, to be discussed later, such a one-
way inhibition would even help to explain the anomalous speed-up in escape be-
havior during positive stimulation. However, the finding applies only to the
lateral hypothalamic rewarding point, as other points have not been tested, and
as mentioned above, it does not seem to comport with the anatomical relations
involved.
Some evidence also exists for inhibition in the other direction when more
anterior reward points are used. Brady and Conrad (2 I) report that normal “fear”
reactions fail to appear in rat and monkey when the animal is working for brain
shocks in some areas near the septal region. These animals, when working for
food or water, will stop and cower if a signal announces oncoming pain-shock. The
same animal working for septal brain shock seems to ignore the danger signal.
The suggestion that septal stimulation might cause relief of pain or fear is in
good accord with reports of clinical investigators that stimulation in septal area
(53) or other anterior locations (I 50) in human beings brings relief and relaxation
PHYSIOLOGICAL INTERACTIONS
Major questions requiring extended work lie in this area and all too few
answers are available. Effects of CNS damage on the positive reinforcement of
hypothalamic stimulation are almost unknown; projections of peripheral stimuli
to the reinforcing centers have not been satisfactorily investigated; and the rami-
fications of the rewarding stimulus on the electrical activity of the rest of the brain
have been studied only in the most cursory fashion.
Effects of CNS Damage

Two studies utilizing surgical ablation and one mentioning electrolytic lesions
in relation to self-stimulation experiments have appeared. An extended study of
the effects of spreading cortical depression of the Bures (30) type on hypothalamic
approach behaviors and tegmental escape has yielded quite striking findings.
The relative unimportance of amygdala, septal area and fornix columns to
the basic phenomenon is indicated in studies of Ward (I 70-172) He implanted
electrodes in rewarding areas in medial forebrain bundle of ventrolateral tegmen-
turn and tested these in one case (I 7 I) after suction ablation of septal region and
fornix columns. Because thirteen rats with septal ablations showed strong teg-
mental self-stimulation, he concluded that septal region and fornix system are not
essential for mediation of basal tegmental self-stimulation. Similarly large lesions
in amygdala were often without influence on tegmental self-stimulation. Ward
(169) suggests that the various areas yielding self-stimulation subserve parallel
rather than mutually prerequisite functions.
Augmentation of lateral hypothalamic self-stimulation by lesions in the “sa-
tiety” center of medial hypothalamus has already been mentioned (6 I).
Functional impairment of the whole cerebral cortex, however, yields quite a
different story. Spreading depression Leao (72) produced by the method of Bures
(29) has caused complete and specific obstruction of the hypothalamic self-stimula-
tion phenomenon (30). The method (29) involves production of spreading corti-
cal depression in a chronic animal by application of KC&soaked pledgets to ex-
posed cortex through trephine openings. With bilateral applications of 22 % KCl,
repeating waves of spreading depression occur which preclude all food-directed
behavior for as much as 4 hours (29).
It is now shown (30) that similar application of KC1 in self-stimulation experi-
ments causes immediate and complete cessation of self-stimulation behavior
driven by lateral hypothalamic stimulation. During the same period, a large compo-
nent of the escape behavior driven by dorsomedial tegmental stimulation con-
tinues. Specifically, it appears that a voluntary or learned component of the escape
behavior disappears with the self-stimulation, and that a reflex or unlearned
component of the escape behavior survives.
The same study presents an interesting account of neurophysiological corre-
lates which may cast light on the mechanisms involved. Unit activity at the dorso-
medial tegmental escape point is briefly augmented during the spreading depression
period; unit activity at the hypothalamic self-stimulation point is greatly depressed.
A corticoreticular inhibitory path which has been physiologically defined in
several studies (e.g., I, 63) would probably account for the great augmentation of
unit responses at the dorsomedial tegmental escape point. No similar
corticohypothalamic facilitatory path has emerged, however, to account for the
depression of unit activity in the lateral hypothalamic self-stimulation area. While
the possibility of such a path should now be taken seriously, an alterna-
tive hypothesis is that the excessive activity at the dorsomedial tegmental escape
point directly inhibits activity in the lateral hypothalamic area. This would
accord well with the inhibitory effects of dorsomedial tegmental stimulation on
hypothalamic self-stimulation (I 29) reported in the previous section.
The hypothesis of such an inhibitory relation would serve yet another explana-
tory function. It might be a means of rendering equivalent the reinforcing event
following hypothalamic self-stimulation and that following the learned tegmental
escape response. In each case there would be augmented activity in the lateral
hypothalamus as the reinforcing event; in the self-stimulation case it would result
from direct stimulation, and in the tegmental case it would be a rebound from
inhibition. From this theoretical basis it has even been suggested that possibly the
learned component of the escape behavior is sustained by the same mechanisms

as the self-stimulation, and therefore both would be expected to disappear together
under spreading depression, as the Bures et al. (30) data demonstrate.
This would also provide some explanation of the anomalous outcome of the
interaction study (129) mentioned previously, in which the escape behavior was
actually augmented by a background of rewarding stimulation in lateral hypo-
thalamus. If the learned escape behavior like the learned self-stimulation is in fact
sustained by lateral hypothalamic activity, then more activity in this region might
be expected to result in more behavior.
Electrophysiological Ramzjications of the Stimulation

Well hidden in the Journal of the Experimental Analysis of Behavior is the out-
standing physiological contribution in this area by Porter and his associates (141,
142). These experimenters used multiple electrodes implanted in monkeys and re-
corded from leads in amygdala, hippocampus, septal areas, and hypothalamus
during self-stimulation tests. They found that o.5-second trains of telencephalic
stimulation in medial forebrain bundle areas caused a spike and slow-wave com-
plex to develop quite often in septal and hypothalamic leads. The complex had a
latency of 25 to 75 msec and a duration of almost 0.4 second. Self-stimulation
tended to cease when the medial forebrain bundle stimulus ceased to elicit this
complex (or when it started to elicit a similar complex in amygdala). With self-
stimulation in the anteromedial nucleus of thalamus, the stimulus produced a
somewhat similar response in septal area and hypothalamus, but in this case there
were also afterdischarges quite identical in appearance. When afterdischarges
appeared, the animal would pause and then self-stimulate again when the dis-
charges came to an end. In this case, the spike and slow-wave complex, whether
during stimulation or during afterdischarge, was often associated with a fast wave
(I 2-15/set) late in each self-stimulation session.
With electrodes in hippocampus, self-stimulation occurred, often in conjunc-
tion with hippocampal seizure activity. In several cases the animals continued to
self-stimulate only so long as seizures were produced, and lost interest if seizure
activity became “adapted out.” With other hippocampal electrodes, self-stimula-
tion occurred at intensity levels well below seizure threshold levels.
In the case of some amygdala electrodes, quite the opposite correlation ap-
peared Animals would self-stimulate only so long as the stimulus failed to induce
seizures. After seizures appeared, for long periods of time, often more than 24
hours, animals would not work.
In the case of posterior hypothalamic and tegmental self-stimulation where
response rates were very high, the continuous stimulus artifact prevented satis-
factory recording. It appears, however, that there were no afterdischarges follow-
ing stimulation in these areas.
In summary, posterior medial forebrain bundle self-stimulation went too fast
for recording; anterior medial forebrain bundle self-stimulation yielded septal
and hypothalamic spike and slow-wave complexes that appeared in some cases
“necessary” to maintain self-stimulation, but there were no afterdischarges; an-
terior thalamic self-stimulation yielded similar waves with afterdischarges, and in

this case, pausesduring the afterdischarge seemedto indicate that it was reward-
ing. Hippocampal self-stimulation yielded full-fledged seizuresthat appeared to be
positively reinforcing because they were in several casesnecessary to continued
self-stimulation. Amygdaloid self-stimulation sometimes yielded seizures which
appeared to be negatively reinforcing because they brought self-stimulation to
an abrupt halt with no resumption after the seizure had subsided.
Interpreting their work, Porter et al. (142) mention the autogenic seizures
in some epileptic children (I 0). In these cases,the patient appears to start his own
seizure by causing a strong-light flicker. In a few casesthis is related to verbal
reports of pleasure. But they also cite Williams (I 74) to suggestthat the emotional
tone of the aura depends on the focus of the seizure.
As their self-stimulation occurred more often than not without even recorded
afterdischarges, the seizure problem is probably not the most important aspect of
their contribution. Rather the single septal-hypothalamic spike and slow-wave
complex which appeared essential to self-stimulation in certain casesmight be
taken as a prototype or perhaps even an exaggeration of a necessarydiffuse elec-
trical event which might characterize the several areas in which crucial conse-
quences occur after medial forebrain bundle reward-stimulation. This might in
the end be most important as a way to identify areas for further and more detailed
study of the physiological consequencesof such stimulation.
Operant Conditioning of Unit Responses

One course which a more detailed study might take is indicated by the
experiments on operant conditioning of single-unit responses(I 22, I 28). These
experiments, which have been underway for about 3 years without yielding
categorical results, were started to determine whether single-unit responsesin
motor or other areas might be treated like peripheral responsesin positive-rein-
forcement tests. In the work already reported, a circuit was arranged to make a
brief train of hypothalamic stimulation occur as a regular consequenceof a given
single-unit responseto learn whether this would cause that responseto increase in
frequency and to occur eventually at a maximum rate.
Rats were prepared first with self-stimulation electrodes in medial forebrain
bundle regions. Preliminary tests established that very high self-stimulation rates
were achieved, and no tendency to escape from stimulation was present. Rats
which failed to meet theserequirements were eliminated. Each rat was then placed
in a stereotaxic instrument under barbiturate anesthesia. A trephine opening of
3 mm in diameter was made in the skull, the dura was cut away and microelec-
trades of I p in diameter (50, 62, I 75) were lowered I mm into the cortex.
As the animal recovered from the barbiturate anesthesia, still in the stereo-
taxic instrument, it was given repeated doses of meprobamate or carisoprodol
(Soma, Wallace Laboratories). The dose of either was 80 mg/kg. It was repeated
whenever any tendency to try to escapefrom the instrument appeared. Previous
tests had shown that an almost paralyzing dose of these drugs fails to block self-
stimulation (I 32). From this point on, the electrode was advanced downward
through the cortex, hippocampal formation, thalamus, and so forth, stopping

whenever a clear spontaneous response appeared. Responses appeared as zoo-
to 500~pv negative spikes, lasting about I msec.
Unit responses were not consid ered satisfactory for these experiments if their
resting frequency was more than about 2 per second; and they were preferred if
this was something less than I per second. When such a response was observed, a
three-step experiment was performed. First, after several minutes of waiting, a
3o-second record was made on moving photographic paper of the spontaneous
rate of the unit response. Second, an elicitation test was made. A series of 20 x-
second trains of stimulation (sine wave 60 cycle/set, 50 pa) was introduced via the
medial forebrain bundle electrodes at a repetition rate of about I every 2 seconds.
In this first test an explicit effort was made to stimulate only in the absence of
single-unit responses. Immediately after this test, a second 3o-second record of
activity was made on film. In the event of elicited effects, each stimulation pro-
duced a series of responses from the unit, and no further tests were made. The
microelectrode was then advanced until a new unit response appeared. If elicited
effects were not found, the experiment continued. Third, a reinforcement test was
made. The experimenter awaited a single-unit response and, each time it appeared,
immediately delivered a stimulus to the hypothalamus. When this was done bY
hand, it was usually applied after each appearance of the response with a delay of
less than s second (this was the experimenter’s response time).
In a successful positive-reinforcement test, the single-unit response rate was
greatly augmented by this reinforcement procedure. The increased rate outlasted
the procedure by a variably period of time. Immediately after this procedure, a
third record of the unit’s activity was made on film. It was the comparison of the
three films that comprised the data.
With microelectrodes in neocortical areas, there were no successful positive-
reinforcement tests by this simple method. With microelectrodes in cingulate
cortex, pyriform cortex, hippocampal formation, and some related subcortical
structures, numerous successful positive-reinforcement tests have been observed.
These took three forms: (z) simple augmentation of response rate of a sporadi-
cally firing unit; (i;) conversion of a sporadic grouped response pattern to a pattern
of continuous firing; and (zX) elicitation of activity immediately after stimulation,
but only when this was given as a reinforcement.
The most striking cases were of the second type. They were recorded mostly
from areas such as the dentate gyrus, fimbria, and mammillothalamic tract, which
appeared to be “seizure-prone” on other tests. In these cases, the unit was origi-
nally responding in a sporadic pattern with single responses or groups appearing
at less than I per second. Stimulation when introduced during silent periods did
not cause any elicited firing. Such stimulation could be continued for periods of
5 minutes or more without materially augmenting response rates. Then, if the
stimulation was withheld and delivered only after the appearance of a single or a
grouped response, 5 to 20 reinforcements would often suffice to cause a sudden
burst of activity; the unit would then respond continuously at rates as high as 30
per second. This burst would sometimes last for a period of only several minutes.
The amplitude of the supposed unit response would often decrease in orderly
fashion during this period. Then the unit response would disappear for a period
of some minutes, to return at the original amplitude and frequency. From this
point on, however, the stimulus even if presented during silent periods would
elicit a burst of responding. It was as if some irreversible change now linked this
response to the area of stimulation.
In other cases, the repetitive activity did not decrease or disappear but con-
tinued at a high level for long periods. In these cases, it appeared that the reinforce-
ment procedure might have made a more or less lasting change in the spontaneous
discharge rate of the unit in question.
The most questionable aspect of these data derives from the anesthetic used;
stimulation of lateral hypothalamus almost definitely counteracts barbiturate and
meprobamate states, and this could account for augmentation in response rate
produced by stimulation. It is also suggested that increments in blood pressure
might easily be involved (M. Mancia, personal communication). The results have
not been regularly reproducible under curarization with Flaxedil. This is not
surprising as animals under pain or stress do not self-stimulate (I 29). But the neces-
sity for a pain-relieving agent which has its effects reversed by stimulation forces
the experiment to rely heavily on the preliminary uncorrelated stimulus control.
If the uncorrelated stimulus fails to augment unit responses but the correlated
stimulus causes augmentation, then the result is clear. However, elicitation does
sometimes occur, perhaps as frequently as reinforcement. Thus the possibility
always remains that in the reinforcing cases, elicitation tests were simply stopped
too soon for the given anesthetic levels. If elicitation tests could be made a second
time after reinforcement tests, the difficulty would be circumvented. But it is
usually not possible to reverse the procedure. This is not surprising as an animal
who has learned and extinguished a lever response for brain shock reward will go
back to responding if given a free series of trains. On the other hand, it makes the
proof of reinforcement as distinct from elicitation all the more difficult.
Four methods are being used presently to circumvent the problem. First is
the search for anatomical distinctions: if it is possible to discover an easily repeat-
able arrangement of stimulating and recording electrodes such that elicitation
never occurs at all, and if reinforcement can be shown regularly with the same
arrangement, then the demonstration will be definite. Second is the double-
barreled recording of experimental and control units from the same general area
at the same time; in this case elicitation applies equally to both units, but reinforce-
ment applies only in the correlated case. Third is the attempt to produce full
reversal of response pattern by use of an automatic stimulator which under reverse
conditions rewards the animal only after a period of no responding. Fourth is
the use of chronic implantation of microelectrodes to dispense with anesthesia
and restraint.
That the experiment has a good likelihood of eventually yielding a definite
proof of operant conditioning of unit responses, and that it will probably be possible
to reinforce unit responses in quite a wide brain area, are suggested by the relative
ease with which the same response can be recorded for very long periods of time
October zg 62 HYPOTHALAMIC SUBSTRATES OF REWARD 579
(sometimesseveral days) if the reinforcement procedure is used. Under no stimu-

lation or uncorrelated stimulation, responsesregularly disappear within the first
hour.
Autonomic Res-onses
Hess (56) divides hypothalamic responses into two types. “Ergotropic”
responsesare related to the sympathetic nervous system but include aswell somatic
expressions.They are considered to enable muscular effort as in defense, attack,
or flight. Among them are pupillary dilatation, rise in blood pressure, increase in
pulse rate, activation of respiration, increase in motor excitability, and general
excitement of the animal.
These are contrasted with “trophotropic” activities of a parasympathetic type.
They releasetension by diminishing the capacity of the organism to produce physi-
cal effort. They provide rest and restitution after strain. They include slowing of
respiration, drop in blood pressure, micturition and defecation, salivation, pupil-
lary contraction, and lossof skeletal muscular tone.
By electric stimulation in chronic cats, Hesshas produced ergotropic responses
in a broad strip of posterior sub- and hypothalamus and anterior midbrain.
Generally it is an area above the mammillary bodies. In lateral sections, however,
it shootsforward in a narrow strip along and parallel to the fornix, almost to the
preoptic area. The mammillary bodies themselves appear to be unresponsive. All
the ergotropic responsestend to go together so that a point which gives one of
these effects ordinarily produces all of them.
Trophotropic or parasympathetic effects are dispersed over the septum, the
preoptic area, and all parts of the sub- and hypothalamus anterior to the ergo-
gropic region. They do not show asmuch tendency to go together as is seenin the
ergotropic responses.Micturition, defecation, slowing of respiration, and decline
in blood pressureare seenon stimulation of parts of the septal area and all through
the anterior hypothalamus back almost as far as the ventromedial nucleus, except
for the narrow area around the fornix, which gives sympathetic responses(or
mixed sympathetic and parasympathetic). The lateral part of this anterior hypo-
thalamic field also produces the other trophotropic responses,salivation, pupillary
constriction, and lossof muscular tone.
The latter response,called “adynamia” by Hess,involves the animal’s sinking
down like an inert mass,without any of the normal adjustments involved in lying
down. The eyes are left open, and the state is quite different from “sleep,” which
Hess seemsto produce by stimulating the region of the thalamic intralaminar
nuclei.
Finally, an area producing only pupillary contraction and arrest of breathing
extends up into the lower and anterior quarter of the thalamus. The only one of the
trophotropic effects found over the whole anterior regions seemsto be arrest of
breathing, which occurs, as stated, on stimulation of septum, all parts of the an-
terior hypothalamus and preoptic area, and in the lower front quarter of the
thalamus.
Some tendency of parasympathetic-like responsesto be found in areas asso-
ciated with positive reinforcement and sympathetic-like responses to be found in

areas associated with negative reinforcement is suggested (I I I). However, it is by
no means clear that a perfect correlation exists. That there should be mixed auto-
nomic responses from a drive-reward system is reasonable considering the widely
variable nature of the instrumental-and-then-consummatory behavior
involved.
Autonomic responses of an even more mixed nature are regularly yielded by
stimulation of the telencephalic areas of positive reinforcement (45, 48, 67, 68).
Moreover, quite often two mutually opposed effects are dependably achieved
from a single point depending on difference in anesthesia, stimulus parameters,
or other factors. Respiration can be inhibited or excited depending on anesthetic
levels with stimulation in cingulate and orbital cortex points (57). Blood pressure
can be raised or lowered by stimulation of the same point, the nature of the effect
varying with changes of anesthesia or stimulus parameters. Activity of stomach
can be started or stopped by stimulation of the same point; in this case a reversal
of background stomach activity is the rule (68).
Similar reversals appear when excitation and inhibition of somatic move-
ments are studied under stimulation of these same points. In unanesthetized or
lightly anesthetized animals, stimulation of telencephalic areas seems to inhibit or
arrest spontaneous movement. In these same animals, after administration of
anesthesia, facilitation of reflex or cortically induced movement is often observed
(68). It will be remembered that, in the posterior hypothalamus where sympathetic
responses are probably the rule, self-stimulation goes together with excitatory
rather than arresting effects even in the unanesthetized animal.
It is perhaps relevant that both positive and negative behavioral mechanisms of
reinforcement involve excitation and inhibition. The animal is often provoked to
activity by “anticipation” of reward, that is, by some conditioned stimulus previ-
ously associated with a reward. The same animal may be pacified by the applica-
tion of the reward stimulus itself. Conversely, with a punishing stimulus, the animal
may be inhibited by anticipation of it, yet provoked into intense activity by its
application. Thus it is not surprising to find reversible excitations and inhibitions
of somatic movement derived from electric stimulation applied to areas that may
form physiological substrates for these mechanisms. The challenge is to specify
the conditions of the two phenomena in the hope that this might further the under-
standing of the actual mechanisms of reinforcement.
Only one study (92) is reported where the same electrodes were tested for
both self-stimulation and autonomic effects. In this case septal electrodes were
used and parasympathetic responses occurred on application of the rewarding
stimulus.
PSYCHOLOGICAL INTERACTIONS
Arousal
The possiblerelation of the arousal system of Magoun (g I) to psychological
mechanisms of drive has been from the beginning a matter of considerable con-
jecture. The postulation of a general emotional variable whose rise, fall, or steady
states control reinforcement mechanisms is a common property of otherwise

divergent psychological theories. Hull (64) proposed that a high emotional state
constitutes negative reinforcement, and that its abrupt decline constitutes positive
reinforcement. Hebb (54) suggested instead that a state of mild emotion fosters
organized behavior and that the absence of emotion or extreme emotion is detri-
mental to this state of organization. The latter notion was extrapolated to the
view that mild emotion constitutes positive reinforcement, and that strong emotion
constitutes negative reinforcement ( I 53). Magoun’s reticular activating system
seemed to provide a physiological substrate for a general emotional mechanism
and thereby added credence to theories of this type (cf 83). Schlosberg (14g), on
the other hand, objected that the dimension running from sleep through alert
attention to extreme tension which appears to be under reticular control is but
one dimension of emotional experience. As another dimension, he mentioned
plesaure-pain which might be under control of different physiological mechanisms.
It is easy to see how this argument led to a study of physiological and anatomi-
cal relations as soon as brain points yielding positive and negative reinforcement
as well as activation were uncovered. Experiments (47, 130) were undertaken to
find whether mild stimulation in RAS might be rewarding, intense stimulation
punishing; or whether points yielding arousal might be identical with or different
from those yielding punishment or reward.
The results indicate that at least three systems can be discovered in the
tegmentum: (i) a system producing negative reinforcement at all electric current
levels and cortical arousal at very low thresholds; (i;> a system producing positive
reinforcement at all current levels but requiring intense stimulation to produce
cortical arousal; and (iz?) possibly a small area which yields neutral arousal with
no positive or negative motivational effects. Specifically, system i occupies a circle
surrounding the middle reticular activating system, system ii occupies the ventral
area below the medial lemniscus, and system iii lies inside the circle formed by
svstem i. Confusing the issue, however, is the fact that inside the circle there are
also points which yield positive and negative reinforcement simultaneously; if
these were perfectly balanced at some point, one might get the appearance of
neutral arousal.
Two possibilities therefore remain. The most likely is that three different physio-
logical substrates exist: one for positive reinforcement, one for negative reinforce-
ment, and a third for physiological arousal. These, of course, are quite likely
related by mutual facilitatory and inhibitory projections as all systems of CNS are.
The alternative still exists, however, that no mechanism of neutral arousal exists,
that all emotion provoking stimuli and all physiological substrates are intrinsically
either positive or negative in emotional tone.
Arrest
The arrest and petit ma1 reactions of Hunter and Jasper (65) have also seemed
related to the reinforcing mechanisms. First, in maze experiments (109) where
positive reinforcement was produced by application of stimulation to septal and
anterior paraolfactory structures, it was observed that application of the reinforcing
582 JAMES OLDS Vohne 42
stimulus in mid-maze caused an abrupt pause. The animal stopped in mid-

behavior for a variable interval depending on the site of stimulation. It was tist
thought that some correlation of arrest and reinforcement might exist but upon
analysis it was found that the anatomical points of longest arrest were also points
of least reinforcement. Later it was found that stimulation in posterior hypo-
thalamic medial forebrain bundle caused no arrest at all.
Porter et al. (142) also noticed the arrest phenomenon in conjunction with
anterior thalamic self-stimulation. In this case, the afterdischarges mentioned
earlier were always accompanied by the petit mal-like arrest reaction.
Many commentators have referred to the autogenic seizures mentioned
earlier (IO), and suggested that this association of petit mal-like states with self-
stimulation is an intrinsic rather than an accidental correlation (I 03, 105, I 42).
Sometimes it is suggested that the petit mal state is relaxing or pleasant (I 03), or
that it provokes automatic response repetition somehow related to a retrograde
amnesia ( I 05).
Considering the large amount of self-stimulation that occurs without arrest or
seizures and the multiplicity of phenomena that appear under the self-stimulation
label (cf p. 559, METHODS OF MEASUREMENT), it appears wise to take a very broad
view. Self-stimulation can occur with or without seizures; and seizures may be
positive or negative in emotional aura depending probably on anatomical locus of
the focal lesion. Occasionally, self-stimulation may be directed toward a “forget-
ting of pains” intrinsic to spreading seizures generally, but this is certainly not
usual.
Perce)km
Perhaps the best answer to the seizure question is the demonstration by Beer
and Valenstein (8) that the animal can be alert and attentive during self-stimula-
tion. These investigators established hypothalamic self-stimulation behavior in
hungry animals which had previously been trained to discriminate between two
tones, one of which signaled the availability of food. The tones were then presented
for discrimination during the actual brain stimulation interval of the self-stimula-
tion test. The tones started after the onset and terminated before the end of the
reinforcing stimulus train. Even under these conditions, animals discriminated the
tones well and stopped to eat when the food-related tone appeared.
Learning
While perception occurs during hypothalamic self-stimulation, it is not at all
clear whether someconfusion of learning or associativemechanismsmay not occur,
at least with the self-stimulation electrode in certain areas. Maze experiments
first indicated that, while rats moved faster for a ventral telencephalic reward,
they still learned more slowly than if the reward were food. Some confusion from
the brain stimulus was suspected (I og).
Stein and Hearst (160) demonstrated quite clearly that the rewarding brain
stimulus in septum or anterior hypothalamus had a severely retarding effect on
acquisition of a discrimination habit if presented during instead of after completion
October Ig 62 HYPOTHALAMIC SUBSTRATES OF REWARD 583
of learning. In this experiment a hungry animal got food by pressing one of two
levers. Auditory stimulusA signaled availability of food at the left lever; auditory
stimulus B signaled food at the right lever. For different rats, rewarding brain stimu-
lation accompanied the onset of one of the two stimuli (A for one rat, B for another).
In each case,the animal learned far more slowly to respond to the stimulus which
was accompanied by the rewarding brain shock. About 75 trials sufficed for perfect
responding when the brain shock was withheld. After 250 trials, responding was
still far from perfect when the brain shock accompanied the auditory stimulus.
It wasshown that the brain shock was more than a simple distraction, asmanip-
ulation of light in the test chamber had no similar effect. The rewarding properties
of the stimulus were implicated by demonstrating that a food reward presented
simultaneously with a discriminative stimulus also caused retardation of learning.
In another study (I 37) a wide sampling of brain points wastested for disrupting
effects of brain shock on a discrimination reversal problem. In this case, after
extensive pretraining, the animal would learn and relearn approximately the
sameproblem day after day in practically the samenumber of trials. Food reward
was given on a contingent basis after the correct response.Brain stimulation was
given on an uncontingent basisin ?&second trains every 3 secondsall during the
test procedure. Points in hypothalamus and rhinencephalic structures caused
disruption. Points in neocortex, primary sensory systems and many tegmental
areas did not. This led to a supposition that disruption was correlated with reward-
ing effects of electric stimulation. The supposition was confirmed by implanting
rats with electrode pairs in the lateral hypothalamic self-stimulation areas and in
the dorsomedial tegmental escape areas. After pretests to confirm elicitation of
strong emotional responses, each animal was tested for learning first during
stimulation of one electrode pair and then for learning during stimulation of the
other. In all cases,animals learned quickly under the negative reinforcing stimu-
lation, and learning was totally disrupted by stimulation at the rewarding point.
It has been suggestedthat thesefindings have more to do with the reinforcing
properties of the stimulus than with any direct relation of the brain points to asso-
ciative mechanisms.In a standard reinforcement experiment reward is applied on
a contingent basis. If the animal makes the correct response, the reward occurs;
otherwise not. In such a case, the reward fosters correct performance. Reward
applied on an uncontingent basis during problem solving has a less certain
status. Several experiments (32) have shown that a distribution of reinforce-
ments on the basis of some ratio between the wrong and the right response
eventually causes a similar distribution of responses. It has been suggested
(H. F. Harlow, personal communication) that this might be spoken of as
partial reinforcement of the wrong response,and that this principle applies to
the caseswhere learning seemsto be inhibited by a rewarding stimulus. The
animals receive the positive reinforcement of brain stimulation for both right and
wrong responseswith the additional reward of food for the right response.This
partial reward of the wrong responseperhaps does account for the animal’s failure
to eliminate the error from their repertory.
The possibleobjections to this simple explanation also need to be voiced. First,
such an explanation certainly could not be applied to recent reports of impairment

of discrimination learning by lesions in some of the same self-stimulation areas
(I 66). Second, it is interesting, if partial reward of wrong responses causes im-
pairment, that partial punishment of right responses by dorsomedial tegmental
stimulation causes no similar impairment. Third, this explanation does not ac-
count for the faster running but slower learning of the early maze experiments
(dog), nor for the confusion caused in the Stein and Hearst (I 60) experiment
where the wrong response was not reinforced at all. These arguments tempt one
to suppose that some real confusion of associational processes occurs as a result of
excessive stimulation in these reinforcing areas.
It is perhaps possible to bring the two lines of explanation together by arguing
first that uncontingent or excessive excitation of positive emotional mechanisms of
the brain has a far more disorganizing effect on choice behavior than similar
excitation of negative emotional mechanisms. On the other hand, the same posi-
tive stimulus applied on a normal and contingent basis causes maze learning, as
was indicated earlier; thus it can cause organization of associative processes. From
these two points it is perhaps not too great a leap of thought to suppose that the
positive-reinforcing mechanisms have far more control over the associative processes
that guide behavior at a choice point than the negative ones have.
Social Interaction
Delgado is reported (22) to have introduced brain stimulation into a monkey
colony. Of particular interest is a taming and possibly reinforcing stimulation of
the dominant male on a prolonged and noncontingent basis. During the period
of stimulation, the position of dominance was usurped by another male, only to
be regained when the electric stimulus was eventually stopped.
PHARMACOLOGICALINTERACTIONS
A comparison of the relative dependability of generalizations in the various
areas of physiology is not particularly favorable to neuropharmacology. Research
experience has given ground for considerable confidence in the reproducibility of
data from homologous areas of mammalian brains; this holds for correlation with
behavioral and physiological functions and appears to be true even across species.
Experience offers no similar ground for confidence in the pharmacological area.
It is common knowledge that doses for one species have no particular relevance
for another. It is becoming clear that even a group of animals within the same
species will often yield a broad family of dose-response functions.
E$ect of Drugs on Electric Stimulation

A series of pharmacological studies of self-stimulation has been made to test the
hypothesis that chemicals which successfully control psychotic agitation may have
in common the property of suppressing self-stimulation. This view was derived
partly from the satiation tests mentioned earlier (I I 3), indicating that animals
with hypothalamic electrodes self-stimulate to exhaustion, continuing for periods
of more than 24 hours. From this uncontrolled responding it appeared possible
that a positive feedback processwas involved; a process,that is, which grows to a

maximum state and continues there. Such a processwould constitute a danger to
the organism because it would trap the animal in unidirectional behavior. This
suggested an alternative to the popular notion that an excess of sympathetic
activity underlies psychotic agitation (24). Many episodesof psychotic agitation
might have quite a different etiology, namely, an excessof the positive feedback
process subserving positive reinforcement mechanisms. A corollary was the hy-
pothesis that chemicals which successfully control psychotic agitation would also
suppressself-stimulation.
Preliminary reports (I IO, I I g, I 24, I 25), indicated that reserpine and chlor-
promazine which have been useful in control of psychotic agitation do suppress
self-stimulation and that pentobarbital and meprobamate, which have little value
in psychosis, do not. It was also indicated that reserpine and chlorpromazine
differed in their effects on self-stimulation depending on electrode sites (I 24, I 25).
This led to the hope that there might be drugs specific to certain drive-reward
systems,a hope which has not yet been realized in experimentation.
Understanding of chlorpromazine action in rats was greatly advanced when
it was compared with meprobamate, pentobarbital, and morphine in combination
approach-escape tests (I 33). Chlorpromazine caused self-stimulation to cease in
2-mg/kg doseswhich permitted escape behavior to continue. Pentobarbital and
meprobamate had just the opposite effect, halting escapebehavior in dosesthat
allowed self-stimulation to continue; the doseswere 20 and 80 mg/kg, respectively.
Morphine fell between, halting self-stimulation and escape behavior at the same
B-mg/kg dose.
The appearance of selective action against positive reinforcement achieved
with chlorpromazine in this experiment was superficially in conflict with other
reports (33,46) that indicate that chlorpromazine acts selectively against avoidance
or defensive mechanisms. When the effects of chlorpromazine on the approach-
escape test and on the other avoidance and defensive reactions were viewed in
more detail, however, the apparent conflict disappeared (I 23). In all cases,chlor-
promazine acted selectively against the voluntary or anticipatory component of
behavior. In almost all negative-reinforcement tests the animal could heed a
warning signal and by somepreparatory responseavoid or diminish the negative
reinforcement. It was avoidance behavior with this anticipatory character that
disappeared under chlorpromazine. All self-stimulation behavior had this anticipa-
tory character, as the animal was never stimulated until after the responseoccur-
red; and all self-stimulation behavior disappeared under chlorpromazine.
In its action on avoidance and self-stimulation mechanisms, chlorpromazine
had almost the same effects as spreading cortical depression (30). It terminated
self-stimulation altogether and also the voluntary or learned component of the
escape response. If the explanation proposed earlier for the action of spreading
depression is valid, namely, that self-stimulation and the learned component of
the tegmental escaperesponse are both sustained by activity of the same lateral
hypothalamic system, then the present data suggest that chlorpromazine acts
selectively to antagonize or raise the thresholds of this system.
Whether this selective action of chlorpromazine against the lateral hypo-

thalamic system is related to its antipsychotic properties cannot be determined
directly because present knowledge of actual mechanisms is too sparse. Indirect
evidence, however, can be drawn from data correlating the two kinds of effect.
Bennett (9) cited evidence supporting the view that prochlorperazine and triflu-
promazine are more efficacious than chlorpromazine against psychotic symptoms;
and that promethazine and promazine are of questionable value in treating these
symptoms. Chlorpromazine itself fell between these two extremes. It is interesting,
therefore, that these phenothiazines were similarly arrayed in their antagonism to
self-stimulation (I 35). In these testswith rat, the 0.5- to 2.0.mg/kg doserange was
used with all compounds. Promethazine augmented self-stimulation. Promazine
had no efficacy at all. Chlorpromazine effectively antagonized self-stimulating at
2.0 mg/kg. Prochlorperazine and trifluoperazine (a triflupromazine-like com-
pound) were extremely efficacious against the behavior even at 0.5 mg/kg.
A seriesof simple but ingenious tests by Stein and others has made it quite
clear that chlorpromazine and possibly other phenothiazines act specifically to
counteract the lateral hypothalamic system by raising thresholds, and that some
antidepressant compounds have an opposite effect, facilitating the same system
by reducing thresholds. One method permitted the animal to indicate its own
threshold (I 58, I 62). Each successivebrain shock was reduced in intensity by a
small step. There were 15 or 20 equal current stepsbetween a moderately reward-
ing top value and zero. A second lever could be operated at any time to reset
current to the top step; but the animal had to take time off from self-stimulation
to do this. The test animal operated the stimulation lever until the current was
driven down to a nonrewarding level and then indicated this level by operating
the reset. In this way the animal by its responsepattern repeatedly indicated the
smallest satisfactory stimulus level and thereby permitted a continuous recording
of these “thresholds” over time.
Chlorpromazine at the low dose of I .5 mg/kg in rat caused a distinct rise in
thresholds with no cessationof responding. More active phenothiazines, proketa-
zine, and trifluoperazine caused self-stimulation to ceasein o.6-mg/kg doses,and
therefore the effect of these on threshold was not definitely established. On the
antidepressant side, amphetamine at 0.75 mg/kg caused a very marked fall in
thresholds, although it was sometimes not quite clear whether animals under
amphetamine showed a willingness to respond without any reward at all. The
barbiturates, pentobarbital at IO mg/kg, and phenobarbital at 30 mg/kg, both
yielded distinct and unquestionable facilitation of the self-stimulation responseby
causing a definite lowering of thresholds.
In a further study (159) involving concurrent self-stimulation at two reward-
ing electrode sites, Stein produced full confirmation of the theorem that chlor-
promazine raises self-stimulation thresholds and that amphetamine compounds
cause these same thresholds to fall. In these concurrent two-pedal, two-electrode
self-stimulation tests, electric current was set considerably above threshold in a
nonpreferred anterior hypothalamic location and just on the verge of threshold
in a preferred, posterior hypothalamic position. The difference between these two
October rg62 HYPOTHALAMIC SUBSTRATES OF REWARD $37
settings causedthe animal to distribute responsesevenly between the two levers. In

this caseany drug which merely activated or quieted without changing thresholds
at the stimulation siteswould have causedequal modification on both levers. But if
a chemical specifically raised thresholds, it would cause a shift away from the
borderline threshold electrode, and if it specifically lowered threshold, it would
cause a shift toward this same electrode which was preferred at suprathreshold
values.
In this situation, metamphetamine proved to have a selective effect in lowering
thresholds, i.e., sensitizing the reward system; and chlorpromazine had the oppo-
site effect, raising thresholds or desensitizing the system.
In a final study, Stein and Seifter (163) analyzed the paradoxical pheno-
thiazine imipramine which appears to counteract psychotic depressionrather than
psychotic agitation when tested clinically. In these studies a self-stimulating rat
was tested first with the electric stimulus just above threshold levels. In this case
both chlorpromazine and imipramine depressedresponding. Then a second test
was made with electricity set below threshold. Under such conditions neither
chlorpromazine nor imipramine was efficacious. Amphetamine applied during
subthreshold tests lowered thresholds and caused self-stimulation to occur. Against
this background of amphetamine-induced responding, chlorpromazine antago-
nized the self-stimulation behavior but imipramine facilitated it greatly. The
authors concluded that amphetamine acted centrally by mimicking the adrenergic
catechol amines which were thought to be the normal stimulators of the reward
system. Chlorpromazine was thought to act centrally by blocking these same
adrenergic mechanisms, and imipramine was said to be efficacious by favorably
influencing adrenergic activity in the reward system.
The view that compounds are selective antagonists or synergistsof the reward
system insofar as they have these samerelation to adrenergic mechanismsis partly
supported by other work comparing chlorpromazine with amphetamine com-
pounds. Miller (97) used the ambivalent responsementioned earlier (I I) to test
chlorpromazine and methamphetamine for effects on reward and escapebehaviors.
He tested with two pedals, one to turn the ambivalent stimulus on, the other to
turn it off. Animals moved regularly back and forth turning the stimulus off and
on. Methamphetamine at 2 mg/kg caused a decline in this shuttling behavior, a
decline which started 45 minutes after injection and lasted for more than 45 min-
utes. Chlorpromazine (4 mg/kg) caused a similar decline which started 15 min-
utes after injection and lasted more than 75 minutes. On the surface the two
depressionslooked similar. When the data were analyzed in terms of speed of the
turn-on and turn-off behaviors, however, a radical difference was shown. Metham-
phetamine caused the turn-off responseto slow with the turn on responseas fast
as ever. Chlorpromazine caused great slowing in the turn-on responsewith the
turn-off responsestill occurring quite rapidly. Thus methamphetamine selectively
depressedthe escapetendency. Chlorpromazine selectively depressedthe reward
behavior. A similar selective action for amphetamine (3 mg/kg) in rats was demon-
strated in another study (I I 9). This showed that the slow self-stimulation produced
with ambivalent electrodes could be transformed to very rapid self-stimulation
by injecting amphetamine. A similar selective action against negative mechanisms

was demonstrated for meprobamate (80 mg/kg).
On the other hand, LSD-25, the violently psychotomimetic compound
which, like amphetamine, may be a monoamine-oxidase inhibitor and therefore
might be expected to facilitate adrenergic mechanisms, caused great increase in
escapebehavior and a decline in self-stimulation when administered at 0.2 mg/kg
(I rg, 125). Another monoamine-oxidase inhibitor is LSD’s close relative, bromo-
LSD (BOL), which has no psychotomimetic properties; some workers think that
it is kept out of the brain because it fails to cross a “blood brain barrier.” On
self-stimulation it sometimesmimicked LSD’s action halting self-stimulation with
electrodes at some places (0.5 mg/kg) (I 16, I 25). But with electrodes at other
places it failed to have any effects.
This effect of BOL was related to that of serotonin. Serotonin is a closerela-
tive of the adrenergic catechol amines; and it is also thought to have excitatory
or inhibitory transmitter action of one sort or another in hypothalamus. It is
implicated antagonistically in many of LSD’s actions outside the CNS, but it
does not cross easily from blood into brain, and is also thought by some to be
stopped by a Cc blood-brain barrier.” In self-stimulation tests, serotonin (0.9
mg/kg) had no effect of its own but it counteracted the LSD effect in the caseof
some electrodes. These were precisely the same electrodes which were unaffected
by BOL. If these were electrodes planted in areaswhere some“barrier” prevented
an action of BOL, then, one might guess,serotonin had caused that same barrier
to prevent the LSD effect.
Direct ChenzicalStimulation
All the preceding experiments have not determined whether the pharmacologi-
cals have direct actions in the hypothalamic self-stimulation system, or yield
metabolites with direct action, or have instead some peripheral or some other
CNS target which in turn causesthe hypothalamic changes.
In an early effort to study direct effects in hypothalamus, chlorpromazine was
injected via cannulae proximal to self-stimulation electrodes. Very large dosesof
more than one-fourth the peripheral requirement were needed to antagonize
self-stimulation (Olds, unpublished). Just the amount of fluid required was enough
to cause great local damage to the area stimulated, and the outcome was therefore
superficially incompatible with the notion of local effects in hypothalamus. How-
ever, it was found that the chemical and electrical stimulus fields did not line up;
the electrical field was spherical below the stimulus tips. The fluid was forced by
the pressureof injection back up the outside of the probe. Thus it formed a cylin-
drical field above the tips.
After this preliminary effort, testsfor direct chemical stimulation were begun,
eliminating the electric probe altogether. In these (‘self-injection” experiments,
pipettes were selectively implanted to deposit chemical substancesin the lateral
hypothalamic reward system. These were connected by polyethylene tubing to a
solenoid-driven syringe. Amounts as small as 1/2000 ml were used. A circuit was
arranged so the animal could drive the syringe by depressinga pedal.
The following chemicals were tested in solutions of I mg/ml of Ringer’s

solution: acetylcholine chloride, adenosine triphosphate, serotonin, adrenaline,
noradrenaline, and iproniazid (I I 8, I 27). Injection contained from 0.5 to I .5 pg
of the chemical being tested. Also, for control, Ringer’s solution was injected alone.
Acetylcholine and serotonin did not cause self-injection behavior in these
dosesbut quickly caused animals to lose muscular tone and apparently to go to
sleep. Adenosine triphosphate and noradrenaline did not have any major effects
in the concentrations used. Adrenaline, on the other hand, appeared to produce
approach behavior; however, it also caused a lossof motor coordination, and self-
injection did not continue for long periods of time. In the caseof iproniazid phos-
phate, there was striking self-injection behavior. Animals injected the chemical at
rates up to 300 self-injections an hour, far above the “chance” rates obtained with
Ringer’s solution alone.
These data were originally taken as supporting the view that adrenaline is an
excitatory transmitter substance for the reward system because iproniazid by
inhibiting monoamine oxidase would be synergistic with adrenergic reactions.
Later, however, it became quite clear that the size of the droplet (I mm3) the
pH of the solution (about 3.5), and the inorganic phosphate used in the salt all
contributed to the excitatory reaction, with no definite evidence of any particular
effect of the iproniazid (Olds, unpublished). Once the pH was adjusted to a stable
neutral level, iproniazid stimulated only on a probabilistic basis; and once the
very acid solution was no longer used, saline in these large dosesstimulated almost
as often.
To eliminate all stimulating properties of the control solution, the droplet
was reduced first to I /IO,OOO ml, and finally to I /IOO,OOO ml. A valve-in-head
probe and pressure-sealeddispensing system were developed to permit these
small calibrated droplets with no diffusion of substance between injections. At
this point no histological damage was caused even by a seriesof several hundred
injections. On the other hand, none of the supposedtransmitters or their relatives
yielded self-injection behavior when the size of the droplet was so small, particu-
larly if the pH were maintained in the 7 to 7.4 neutral range. Epinephrine, sero-
tonin, acetylcholine, and iproniazid were tested in 10-5 mg/ml concentrations;
all were slightly active at a pH of 6; none were active at a pH of 7.
However, dependable self-injection behavior was produced with these small
droplets when certain buffer acids and Krebs cycle intermediates were used (I 36).
Most of these had in common an affinity for calcium. Usually an iso-osmolar
solution of the sodium salt was the basis for determining these concentrations,
although modifications had to be made to bring the pH into the neutral range.
In certain experiments, however, solutions of these concentrations were diluted
to 1/2 and I /4 with NaCl to determine concentration thresholds, and in other
cases tris (hydroxymethyl) aminomethane or K was substituted for Na as the
cation.
The anions tested included inorganic phosphate, citrate, succinate, oxalate,
and pyrophosphate. All these chemicals had some transmitter-like characteristics
insofar as there was a brief period of stimulation after each self-injection, followed
by someprocessof neutralization so that stimulation ceasedand the animal came

back for another dose. These chemical reward stimuli were capable of maintaining
continuous self-injection behavior for long periods of time, the behavior stopping
within several minutes after the withdrawal of the chemical reward stimulus.
The strength of the chemical stimulus as demonstrated either by threshold
measurementsor rate of responsewas directly related to affinity of the test ion for
calcium. Thus, inorganic phosphate was a moderate stimulus, citrate and succinate
were stronger, and pyrophosphate and oxalate were very strong. These substances,
therefore, are assumedto act by withdrawing ionic calcium from the interstitial
fluid, thereby permitting ceils to discharge spontaneously; such a reaction has
been reported often for excised nerve preparations in Ringer’s solution (e.g., 84).
Contrary to initial expectations, potassium not only failed to yield satisfactory
results as a stimulus itself but also inhibited the anionic reactions. Thus, sodium
phosphate caused dependable self-injection behavior; no similar response was
provoked by potassium phosphate. Detailed study of the time relations involved
make it clear that this effect of potassium was to be expected on the basisof earlier
peripheral nerve studies. Potassium regularly stimulates for a brief initial period
after which depolarization goesto extremes and no further dischargesare observed
(84). In self-injection tests, I o minutes of exploratory behavior are involved before
the animal has learned to self-inject; by this time it would be expected that the
stimulatory effects of potassium would be passedand the effects of extreme de-
polarization would have set it. Stimulation of peripheral nerve caused by lowering
ionic calcium levels showsno similar tendency to passover into extreme depolari-
zation (84). This is possibly one reason why self-injection behavior is regularly
produced by substanceswhich tie up calcium.
Several interpretations of the failure of the supposedtransmitters are possible.
First, of course, is the likelihood that satisfactory chemicals or concentrations have
not yet been tested. Second, the end foot might effect so tight a juncture with the
cell that the “chemically sensitive” subsynaptic membrane (cf 52) cannot be
reached by fluids injected into interstitial space. In spite of these highly likely
possibilities of a basically technical nature, it is interesting to speculate on more
theoretically significant possibilities that might be indicated should the technical
explanations eventually fail.
One is the possibility suggestedearlier (cf the previous section on PUNISH-
MENT AND REWARD) that in this group of systems, spontaneous activity is the
rule, synaptic relations being mainly inhibitory. In such a case,injection of inhibi-
tory substancesmight cause changes in behavior, but self-injection would not be
provoked by use of excitatory transmitters. Physiologically, the activity in these
systemswould be mainly controlled by the inhibitory input; however, it is also
reasonable to supposethat the level of spontaneous activity in a given group of
cells might be modulated by physiologically meaningful changes in regional
ionic distribution.
Another possibility cannot be completely ignored, namely, that the simpler
ionic messengersmight actually function as transmitters in the central nervous
system, behind the protective wall of the “blood-brain barrier,” in a fashion which
would be unsuitable in peripheral ganglia or neuromyal junctions where tissues

quickly take on concentration of thesesubstancesfrom the blood. For a transmitter
to function as such, it must be introduced into the synapse by the tierent fiber
only; it cannot be entering regularly by diffusion from other sources. Otherwise
the postsynaptic unit would be firing all the time. As an example of a possible
ionic transmitter system, a case could be made for the theory that inorganic
phosphate yielded by the ATP-ADP reaction in the active end foot would bind
calcium in the interstitial space, causing thereby an excitatory postsynaptic
response. The firing cell might modify pH slightly, causing calcium to return to
the medium and making the inorganic phosphate available again to complete
the cycle by participation in the ADP-ATP reaction.
Other work utilizing different end points has relevance to the question of
brain-stem transmitters and should be mentioned before the subject is dismissed.
First there is a study by Curtis et al. (34) with glassmicroelectrodes used simul-
taneously for chemical stimulation and electrical unit recording in tegmentum.
This seemedto support the view that the supposedtransmitter systemsare without
excitatory effect in somebrain stem systems.In this case, the anion glutamate had
excitatory effect; it is not supposed to stimulate like phosphate by depleting
ionic calcium. The work of Grossman (51), on the other hand, seemedto indicate
that the peripheral transmitters might have somefunction in the lateral hypothala-
mus. In this case, crystalline probes were lowered into the lateral hypothalamic
region. Probes could be withdrawn from the brain to have the crystalline sub-
stanceschanged and then replaced at the samebrain site. The results showed that,
in the sameanimal and in the samehypothalamic location, epinephrine produced
eating behavior, and carbachol, an acetylcholine synergist, produced drinking.
This seemedto indicate that roughly the same hypothalamic area is occupied by
an adrenergic eating mechanism and a cholinergic drinking one. The possibility
that osmotic balance or pH may have been differentially affected by the two
crystalline substanceswas to some degree handled by controls. Therefore the
possibility of functional transmitter systemsin hypothalamus related to the well-
known peripheral transmitters cannot be ruled out.
The current stage of the self-injection experiments involves a return to the
original question of whether tranquilizers act directly on hypothalamic centers.
Against a background of anionic self-injection behavior, intracranial applications
of various compounds were tested. In these preliminary studies, solutions of sodium
pyrophosphate plus an inert ion were used to provide a strong and dependable
chemical stimulus. Then, in this solution, various possible inhibitors were sub-
stituted for the inert substanceto learn whether any inhibition of the self-injection
behavior could thereby be produced. Serotonin, gamma aminobutyric acid,
chlorpromazine, and other compounds were tested in o. I to 0.2 molar concentra-
tions; the three mentioned chemicals all gave preliminary indication of local
inhibition of the anionic self-injection behavior. However, because of special pH
and solubility problems introduced by the use of these mixtures, it is not yet
possible to make a definitive statement about local inhibitors (I 36).
592 JAMES OLDS Vohe 42
SPECULATIONS
Appetitive Behavior and the Olfactory Forebrain
It is not impossible that aversive reactions are the only ones deriving from
the irritability inherent in protoplasm, and that appetitive reactions are a later
phylogenetic development awaiting the evolution of an olfactory apparatus. A
specialized chemosensitive receptor occurs in coelenterates and platyhelminthes;
in both, it serves to guide behavior in pursuit of food. In insects it has come to
subserve another appetite as well, indicated best perhaps by the gypsy moth who
will home over 2 miles on the odor of a gypsy female (66).
The forebrain, as soon as it appears in the phylogenetic series,is also linked
to olfaction. It appears tist in early vertebrates where, in shark for example, the
brain seemsbut a looseconjunction of midbrain and a long forward protuberance,
the olfactory bulb and olfactory lobe. Here again, there is an appetitive function
linked to this olfactory apparatus and its form suggeststhat there is much in
common between our notions of appetitive reaction and “operant” behavior:
amputation of the forebrain causesfish to lose “initiative,” that is, “the ability to
react to stimuli in a specific, nonreflex manner” (I 43).
The experiments of the present review might be taken to indicate that the
appetitive behavior which was from the start wedded to the chemoreceptor system
is still so bound, although the direction of chemosensitivity has shifted away from
its original external orientation and toward the blood and cerebrospinal fluid of
the milieu interne. In the course of this change, what started asappetitive behavior
has evolved into a whole system of operant or voluntary mechanisms and the
forebrain which started as a small olfactory appendage has developed to a point
where it comprises almost the whole brain.
Appetitive Reactions and Drives
In any event, a large system of the brain phylogenetically derived from the
olfactory apparatus and possibly still specialized to chemoreception apparently
functions primarily to mediate conservative and appetitive reactions. The former
are uncovered by observing the autonomic responsesproduced by electric stimu-
lation in these areas, the latter by observing instrumental or consummatory
responsesyielded by the samestimulation. Besidestheseelicited effects, stimulation
of the sameareas has, on random behavior, the effect of a primary reward, causing
avid repetition of those responsesequenceswhich are sufficiently often followed by
the brain stimulus. These primary rewarding effects are demonstrated in self-
stimulation experiments.
Experiments in which basic hunger or sex drives are manipulated during
self-stimulation behavior suggestthat the system is differentiated into subsystems
on the basis of the different basic drives. Other data (5) suggest that chemicals
in the blood related to a particular basic drive constitute a major pathway of
control over the correlated reward subsystem.
Another pathway of control is demonstrated by negative reinforcement
experiments which indicate that certain negative mechanisms have an inhibitory
FIG. 2. Possible organization of the
reinforcement mechanism. Sensory fibers
of paleocortex receive projections from
olfactory, gustatory, and visceral receptors
and project inhibitory impulses (i> onto
drive centers of medial hypothalamus
(VMH). This releases lateral interstitial
fibers @ from inhibition, causing facilita-
tion (f) in medial forebrain bundle (MF’).
Extrapyramidal fibers of origin in paleo-
cortex, caudate, globus pallidus, zona in-
certa, and substantia nigra are shown as
Latera I
System
coursing through the lateral hypothalamic
LHA area (LHA) and related areas which are
MFB occupied by the medial forebrain bundle.
It is supposed that these are the systems
which give direction to behavior and which
- Autonomic are reinforced by the lateral interstitial
fibers. The drive centers of medial hypo-
Consummatory thalamus are supposed to yield both aver-
1 Skeletal
sive consequences
relation
The suggestion
to the lateral
and inhibitory
interstitial
that some part of this area
effects in
system.
is also a glucose receptor shows that some of the aversive activity is thought to correlate with an
excessively satiated rather than an excessively hungry state so far as the food system is concerned.
It is interesting, but not quite anomalous that some medial hypothalamic aversive responses have
been shown to correlate with an excess of satiety (61) ; psychology has tended to emphasize those
aversive mechanisms associated instead with an excess of hunger. As far as the autonomic responses
are concerned, sympathetic activity might be caused by activity in the VMH drive center, and
parasympathetic activity (iJ might be released from inhibition when this drive center activity
subsides. Consummatory responses @ would be similarly released but only on further reduction
of VMH activity. In operant conditioning, the facilitatory relation (f) would directly increase
later response frequencies, and the temporary connection (t) would cause a further indirect incre-
ment in response frequency.
relation to the appetitive ones. The inhibitory relation of the negative feeding
mechanism of ventromedial hypothalamus to the “feeding center” of lateral
hypothalamus, and the apparent subservience of this negative mechanism to
glucose level (5) seem to indicate that drive control over appetitive mechanisms
may itself be mediated via negative areas which are inhibitory in relation to the
appetitive system and negatively reinforcing in relation to behavior (see Fig. 2).
A third line of control is suggestedby the fact that the brain areas yielding
self-stimulation are supposed to be actual substrates of reward. If this is true,
receptors in the viscera and the periphery normally receptive to primary-reward
stimuli must in one way or another send projections here.
In any event it is clear that stimulation of the samelateral area has two usually
dissociated effects. On the one hand, it has the effects of the primary drive itself,
causing emissionof drive-related instrumental and consummatory responseswhen
suitable opportunities are offered. On the other hand, it hasthe effect of the primary
reward related to that drive, causing repetition of the preceding behavior when it is
used to reinforce operant responding. Therefore, the possibility that the electric
stimulus constitutes a simple internal surrogate for either is unlikely. It seems

quite possiblethat neural excitation attributable to primary drive and that attribut-
able to primary reward are both projected to the samearea with subtle differences
in function, and that the electric stimulus, being something of a bludgeon, has
the effect of both at the sametime.
One is tempted to speculate that drive would ordinarily lower thresholds of
instrumental and consummatory responsesrelated to the area, but cause actual
discharge only in autonomic efferents. Primary rewarding stimuli, finding the
thresholds lowered, would then cause discharges in the efferents which control
consummatory responsesand by this same action cause some neural substrate of
the preceding instrumental response to become related to this particular drive-
reward focus.
These relations are schematized in Fig. 2. Drive in this diagram influences
the system when low glucoselevels cause a decline in the activity of the inhibitory
glucose receptor. By suspending inhibition, this causes immediate discharge in
autonomic effecters and lowered thresholds in the other neurons of the lateral
system. Olfactory or visceral afferents then cause discharge in these other neurons,
thereby causing consummatory responsesand somehowtying the preceding skeletal
responseinto this drive system. As for the pathway of control of olfactory or vis-
ceral afferents, the guessportrayed in Fig. 2 is that they further inhibit the medial
inhibitory system thereby yielding spontaneous activity even in the less active
cells of the lateral system.
Mechanisms of Reward
The basic question posed by these findings and speculations is: what does it
mean for the substrate of some instrumental response to become related to a
particular drive-reward focus? One can only speculate. From the experimental
analysis of behavior, someaspectsof the meaning may be guessed,namely, (i) that
its threshold is generally lowered by the relationship so the responsemight be more
frequent in the future, and (ii) that its threshold is also brought into somerelation
with the correlated drive so that the active drive may cause even further lowering
of its thresholds. Possibly also a third consequenceis involved, (iii) that the neural
substrate of the behavior represents a new pathway of control over the focus in
question so that stimuli tending to arouse the behavioral substrate will also have
sometendency to arouse the drive in the future (cf I I 7).
For further hints about the mechanism, we may turn our attention briefly
to the nature of the neuroanatomical substrate of the drive-reward focus. We can
never be sure which of the structures near an electrode tip is yielding a particular
effect. However, if a large number of brain points is tested, and those yielding an
effect follow a patterned course through the brain, it becomes a matter of ever-
increasing likelihood that any anatomical structure following a similar course is
importantly related to the effect. On the basisof such reasoning, it may be guessed
that the main substrate of reward is a set of interstitial elements which forms a
system through the lateral hypothalamus, basal ganglia, and paleocortex. In-
creased activity in these elements appears to be the final behavior mediator of
October 19 62 HYPOTHALAMIC SUBSTRATES OF REWARD 595
the reinforcing effect. An increment of excitation of these elements following

closely after a randomly emitted operant responsemay cause the later repetition
of the operant. On the basis of evidence from pathological tissue, Papez (140)
speculated that these are granule cells and chemoeffectors, that is neurosecretors,
which might accomplish their reinforcing effect by facilitating excitation in their
longer-axoned neighbors, or in passing fibers whose cell bodies might be at some
distance.
From the self-injection experiments it might be guessedthat these interstitial
elements do not have excitatory afferents but instead have high spontaneous
activity rates controlled on the one hand by local ionic balances and on the other
by inhibitory inputs from negative areas. If this were so, the consequencesof
electric stimulation at positive and negative reinforcing centers would be explained.
For, insofar asincrements in activity of these elements causepositive reinforcement
in ongoing behavior, it is clear that stimulation of these cells directly would yield
positive reinforcement at the onset of the stimulus, and stimulation of the inhibitory
afferents would yield similar reinforcement upon termination of the stimulus.
If the interstital elements of the medial forebrain bundle are spontaneously
active, and controlled by inhibition only, and if they form a substrate of reward,
the interesting possibility arises that primary rewards have their effect on the
system by a processof double inhibition. They might be projected to the positively
reinforcing paleocortical areas such asthe entorhinal one, which might be thought
from the work of Adey (I) and Fonberg and Delgado (44) to inhibit the inhibitors
of the interstitial elements. Certainly clear evidence of such a pathway exists:
(i) stimulation of entorhinal area causesmoderate reinforcement, aswith a reward
stimulus from the environment (I 8); (z?) stimulation of entorhinal also causes
inhibition of unit responsesin dorsomedial tegmentum (I); (zX) stimulation of
dorsomedial tegmentum causesnegative reinforcement ( I 30) ; and (iv) stimulation
of this same tegmental area causesinhibition of the lateral hypothalamic self-
stimulation response(I 29).
SUMMARY
In summarizing such material as this, it does no harm to make a clear separa-

tion between the establishedfacts and the tempting speculations. The speculations
have been presented. The facts are listed below, grouped according to the major
parts of this review.
I) Electrical stimulation in a very broad set of brain areas yields effects on
behavior tantamount to those of primary reward. The areas involved are largely
in hypothalamus and rhinencephalon.
2) With current correctly adjusted and the electrodes correctly placed, it is
possible to generate more motive force with this type of reward than with any
other reward usedin animal experimentation. With current set lower, or electrodes
differently placed, far milder effects are achieved, effects comparable in every way
with conventional rewards. With telencephalic electrodes, satiation occurs so
that animals will self-stimulate daily for fixed periods of time but not indefinitely.
With somehypothalamic electrodes, there is no satiation so that animals self-stimu-
late to the point of exhaustion.
3) With electrodes in someplaces, the tendency to self-stimulate is a monotonic

function of the electric current level: in these casesthe tendency to approach
becomesmore and more augmented as the current is raised; tests have been made
with current at more than 20 times the threshold settings. With electrodes in
other places, the self-stimulation rate riseswith early current increasesand declines
with later ones.
4) The phenomenon is regularly provoked by stimulation of approximately
the same areas as those previously implicated in studies of various basic drives.
Self-stimulation via differently placed electrodes is sensitive to manipulation of
different basic drives : furthermore, these self-stimulation electrodes often yield
the correlated consummatory responsesif the stimulus is delivered by the experi-
menter and the responsesopportunity exists. Conversely, electrodes yielding con-
summatory responsesoften yield self-stimulation.
5) Besides the areas where stimulation yields primarily rewarding effects,
there are other areas in the brain where stimulation yields primarily punishing
effects, and there is a third and perhaps most numerous set of placements where
stimulation seemsto yield both effects about equally. Medial forebrain bundle
points of telencephalon, diencephalon, and mesencephalon are especially apt to
yield pure positive reinforcement; nonspecific thalamic points, dorsomedial
tegmental points, and points above the medial lemniscus are especially apt to
yield pure negative reinforcement; and points in middle hypothalamus regularly
yield both effects.
There is interaction between some negative-reinforcing areas and some
positive-reinforcing areas such that stimulation of the negative area is antagonistic
to the self-stimulation behavior produced by stimulation in the positive area, but
stimulation of the samepositive area is synergistic to the escapebehavior produced
by stimulation in the negative area. It is not clear whether this is a general rule
for interaction or applies only to the tested locations.
6) Lesions in amygdala or septal area and fornix column which include
certain anterior portions of the positive-reinforcement system do not prevent self-
stimulation via mesencephalic electrodes. Spreading depression in the neocortex
of the rat, on the other hand, causescessation of all self-stimulation and also of
the operant component of negatively reinforced behavior, while reflex escape
behavior survives. The same spreading depression causesvast augmentation in
unit firing at the tegmental escapepoint and vast depressionof unit firing at the
lateral hypothalamic self-stimulation point.
Electroencephalographic studies show that self-stimulation in the medial
forebrain bundle of telencephalon is often accompanied by afterdischarges in
septal area and hypothalamus; it is not clear whether similar afterdischarges occur
with other self-stimulation electrodes, or whether they are actually involved in the
reinforcing process.
With hippocampal self-stimulation, seizures evoked by the self-stimulation
electrode appear synergistic to the positive reinforcing process; with amygdaloid
self-stimulation, seizures evoked by the self-stimulation electrode appear antag-
onistic to the positive reinforcing process. It is clear that seizures are in no way
prerequisite to self-stimulation.
Microelectrode studies indicate that paleocortical single-unit responseswhich

are used to trigger rewarding hypothalamic stimuli often become vastly augmented
in spontaneousdischarge rate. No similar modification in the rate of the triggering
unit has been produced when neocortical units are used to trigger the same
stimulus.
Autonomic responsesare regularly evoked by stimulation of the areasinvolved
in self-stimulation, but it is not definite that these are primarily parasympathetic,
aswas originally thought.
7) The self-stimulating animal is sufficiently alert during the period of stimula-
tion to discriminate between two different tones. On the other hand, uncontingent
application of the stimulus during the learning of a discrimination problem causes
drastic impairment. Similar uncontingent application of a stimulus causesthe
dominant male of a monkey colony to lose his position.
8) Tranquilizers of the phenothiazine group regularly abolish self-stimulation
behavior and the voluntary component of escapebehavior permitting reflex escape
behavior to continue. These effects are almost definitely achieved by a direct or
mediated action of the drugs on hypothalamic thresholds. Barbiturates and
meprobamate do not have similar effects. Activators of the amphetamine group
appear to have just the opposite effect, augmenting self-stimulation by lowering
hypothalamic thresholds.
Microinjection studies show that normal excitatory transmitters applied
directly to hypothalamus do not arouse the reinforcement mechanism; however,
chemicals such as phosphate which causesa loss of local ionic calcium regularly
do arouse the mechanism and can be substituted for electricity in self-stimulation
tests. An extension of this technique is being used to test for direct inhibitory
action of phenothiazines in hypothalamus.
CONCLUSIONS
It is by no means clear from the present data what actual physiological path-
ways mediate the effect of primary rewarding stimuli on the self-stimulation areas.
It is not even proven that these pathways exist. Until such a demonstration is
made, the theory that this system constitutes a veritable substrate for rewarding
mechanismswill remain speculative. Furthermore, if this is a substrate, and I am
quite sure it is, the pathway needs to be defined both for the purpose of clarifying
the mechanismsinvolved and to render the knowledge of these mechanismsuseful
in relation to pathological conditions. Therefore, a next step should involve
physiological and behavioral studiesaimed at finding and defining these pathways.
On a prima facie basis, one would expect afferent pathways from lower
centers and from olfactory and gustatory centers to converge on the system.
Becausehypothalamic lesionscauselossof drive control over the system (I 64, I 65),
and amygdaloid lesionscause a lossof head receptor control over the system (TO),
it is possibleto speculate that drive-related afferents have their accessto the system
via the hypothalamus, gustatory, and olfactory afferents via the amygdaloid
complex. Visceral afferents stand in a middle ground where guessingis difficult-
In any event, a great deal of work is needed as no clear demonstration has been
made of a neurophysiological response at a rewarding electrode site caused by

presentation of a primary reward, or any related stimulus, to the animal.
Similarly, no good demonstration has been made of the pathways from the
hypothalamus that are actually involved when stimulation of the lateral area
serves as a reward to control behavior. Electrophysiological ramifications in terms
of afterdischarges and seizures have been observed in correlation with reinforcing
stimulation in some areas, but it is hard to evaluate whether these are functional
in the reinforcement process or mere side effects. Afterdischarges in the septal
area seemed functional in relation to telencephalic self-stimulation in the Porter
et al. (I 42) study; however, Ward (169) was able to demonstrate mesencephalic
self-stimulation after lesions had destroyed both septal area and the columns of
the fornix. What appears to be needed is a set of tandem studies alternating
between electrophysiological recording and the placing of lesions until those
pathways are located, the destruction of which selectively abolishesthe reinforcing
effect.
Finally, the most important gaps in present knowledge of the self-stimulation
phenomenon stem from insufficient interaction of the newer stimulation-and-
behavior techniques with the classical techniques of electrophysiological research.
GLOSSARY~
“Acquisition” is synonymous with “operant conditioning”; it is opposed to “extinction.”

During “acquisition” the “reinforcement” is presented; during “extinction” reinforcement is
withheld.
“Ambivalent” is an adjective applied to a stimulus which is simultaneously a reward and a
punishment. A reward or punishment which is not ambivalent will be spoken of as “pure.”
“Appetitive behavior” * is behavior reinforced by presenting a previously withheld stimulus
after a response; it is synonymous with “approach behavior.”
“Approach” is synonymous with “appetitive behavior.”
“Arousal” is used here with the many connotations it has acquired from the experiments
on the “reticular activating system” (91).
“Arrest” is a term borrowed from Hunter and Jasper (65); it refers to a sudden stop of overt
behavior.
“Aversive behavior” * is behavior reinforced by terminating a noxious stimulus after a
response. It is synonymous with “escape behavior.”
“Behavior” as used here is any measurable and repeated output from the animal. Usually
it is skeletal but it would apply to autonomic processes and electrical behavior recorded from the
brain. The term would not apply to some outputs as, for example, blood from the bleeding animal;
but I believe my meaning is clear. “Behavior” is synonymous with “response” and “reaction.”
“Drive” * is a condition created by deprivation of food or water or sexual stimulation or
by other privations. According to theory, it directly causes certain reflexes related to the withheld
stimulus; and it also converts the withheld stimulus into a reward. It is specific to the withheld
stimulus so there is a “food drive,” a ‘<water drive,” etc.; these terms are synonymous with “hun-
ger, ” “thirst,” etc. Such drives are often termed the “basic drives.”
“Electrode” actually refers to the uninsulated interface between the electric conductor
and the brain; in most cases it is the uninsulated tip of a wire. When two wires with tips very close
together are used for bipolar stimulation of one point, it is spoken of as an “electrode pair.”
“Elicitation” is the function of an afferent stimulus in relation to the reflex which it causes;
2 Terms followed by an asterisk are considered basic terms of the vocabulary used in this
PaPer and might be studied in advance by those unfamiliar with the field.
October zg62 HYPOTHALAMIC SUBSTRATES OF REWARD 599
the “eliciting function” of a stimulus is distinguished from its “reinforcing function.” “Elicited
responses” come immediately after the stimulus; whereas a stimulus “reinforces” those responses
which immediately precede the stimulus (cf Skinner, I 56).
“Ergotropic” is a term borrowed from Hess (55) to refer to sympathetic responses of the
autonomic variety together with related “voluntary” responses.
“ESB” is an abbreviation meaning electric shock to the brain, and referring usually to a
local electric stimulus delivered via an implanted electrode.
“Escape” is synonymous with “aversive behavior.”
“Extinction” is the opposite of “operant conditioning,” a response is permitted to occur
without “reinforcement” and its rate eventually returns to the “operant level” (cf 156). “Extinc-
tion” also refers to the decline in response rate caused by the “extinction” procedure.
“Motive” is an inclusive term applying to many factors which control operant behavior
such as “drives,” “rewards,” and “punishments.”
“Negative reinforcement” is synonymous with “punishment.”
‘cOperant behavior” * is responding whose frequency is controlled mainly by ‘ ‘reinforce-
ment” from succeeding stimuli rather than by cCelicitation” from preceding stimuli. The term is
synonymous with “voluntary’, and opposed to “reflex,’ (cf 156).
“Operant conditioning” is the modification of operant behavior frequencies by means of
“reinforcement”; it is synonymous with c‘instrumental conditioning” and opposed to ‘cPavlovian”
or “classical” conditioning (cf I 56).
“Operant rate” refers to the initial or random rate of a particular “operant behavior,‘;
before “operant conditioning,’ the rate is “operant”; it is “operant” again after a long period of
“extinction.”
“Positive reinforcement” is synonymous with “reward” (cf 156).
“Primary reinforcement” * is synonymous with “reinforcement”; it is unlearned as distin-
guished from “secondary reinforcement” which is “learned.”
“Primary reward” refers to “positive reinforcement” or 9eward” but implies specifically
a relation of the ccreward” in question to one of the “basic drives.” A “primary reward” is a
strong reward and no learning was required to make it so.
“Punishment”* is noxious stimulation whose termination “reinforces” “aversive behavior.‘,
It is synonymous with “negative reinforcement.”
“Pure” is an adjective applied to a “reward” or “punishment” when the stimulus involved
does not have “ambivalent” effects. The adjective is also applied to “appetitive” or “aversive,,
behavior when the stimulus controlling the behavior is not “ambivalent.”
“Reaction” is synonymous with “behavior.”
“Reinforcement,’ * refers both to any stimulus which is manipulated after a response to
modify its later repetition frequency, and to the process of applying a stimulus after a response to
modify its frequency. The stimulus is called a “reinforcement”; and the process is the process of
“reinforcement.” As a process, “reinforcement” is synonymous with “operant conditioning,” and
it is opposed to “extinction, (cf I 56).
“Response” as used here is synonymous with behavior. A response may be caused directly
by a stimulus, or it may be spontaneous so far as present usage goes. The same is true of the
synonym “reaction”; and one might say with justice that it would be better to say “action” than
“reaction’, in these cases.
“Reward” * is the term used to designate a previously withheld stimulus presented to rein-
force appetitive behavior; it is synonymous with “positive reinforcement” and with “gratifica-
tion.”
“Secondary reinforcement” * usually refers to a stimulus which functions like a “reward”
but which is not related to any of the “basic drives.” It is presumed to have its reinforcing power
by virtue of an association with a “reward”; thus it is a “learned reward.”
“Self-stimulation” is “appetitive behavior” reinforced by ESB. Usually it is the pedal
responding of an animal who is stimulating his hypothalamus (or some related structure) by
means of each pedal response.
“Single-unit response” refers to a repetitive electrical event of 0.~ to I.+msec duration
600 JAMES OLDS VoZum 42
with constant amplitude and template characteristics, recorded from brain by use of microelec-
trodes of I to I o p in diameter, disappearing on movement of the microelectrode by over 40- to
100-p distance. While constant amplitude is supposedly a criterion of a “single-unit response,”
sometimes responses with amplitude decreasing in orderly fashion are accepted as “single-unit
responses. ” “Single-unit responses” are thought to be the propagated action potentials of single
neurons.
“Trophotropic” is a term borrowed from Hess (55) to refer to parasympathetic responses
of the autonomic variety together with related ccvoluntary” responses.
ccVoluntary” as used here is synonymous with “operant.” It is an adjective applied to
behavior which is controlled more by previous consequences than by present eliciting conditions
(cf 156).
REFERENCES
I. ADEY, W. R. Organization of the rhinencephalon. In: I 6. BRADY, J. V. Emotional behavior. In: Handbook of
Reticular Formation of the Brain, edited by H. Jasper Physiology, edited by J. Field et al. Washington, D. C.:
et al. Boston: Little, Brown & Co., I 958, p. 62 I. American Physiological Society, 1960, sect. I: Neuro-
2. ANAND, B. K., AND J. R. BROBECK. Hypothala- physiology, vol. m, p. 1529.
mic control of food intake in rats and cats. Yde J. 17. BRADY, J. V. Temporal and emotional effects re-
Biol. Med. 24: I 23, 1951. lated to intracranial electrical self-stimulation. In:
3. ANAND, B. K., AND J. R. BROBECK. Localization EZectricaZ Studies on the Unancsthetized Brain, edited by
of a “feeding center” in the hypothalamus of the E. R. Ramey and D. S. O’Doherty. New York:
rat. Proc. Sot. Ex#Z. BioZ. Med. 77 : 323, 195 I. Paul B. Hoeber, Inc., 1960.
4. ANAND, B. K., AND J. R. BROBECK. Food intake 18. BRADY, J. V. Motivational-emotional factors and
and spontaneous activity of rats with lesions in the intracranial self-stimulation. In: EZectricaZ StimuZation
amygdaloid nuclei. J. Neurophysiol. I 5 : 42 I, I 952. of the Brain, edited by D. Scheer. Austin: University
5. ANAND, B. K., S. DUA, AND B. SINGH. Electrical of Texas Press, I 961, p. 413.
activity of the hypothalamic “feeding centers” under rg. BRADY, J. V., J. BOREN, D. CONRAD, AND M.
the effect of changes in blood chemistry. EZectroen- SIDMAN. The effect of food and water deprivation
ccphalog. and Clin. Neurophysiol. I 3 : 54, I 961. upon intra-cranial self-stimulation. J. Camp. and
6. ANDERSSON, B. The effect of injections of hyper- Physiol. Psychol. 50: 134, 1957.
tonic NaCl solutions into different parts of the hypo- 20. BRADY, J. V., MD D. CONRAD. Some effects of
thalamus of goats. Acta Physid. Stand. 28 : I 88, I 953. brain stimulation on timing behavior. J. ExptZ.
7. ANDERSSON, B., P. A. JEWELL, AND S. LARS- Analysis Behavior 3 : 93, I 960.
SON. An appraisal of the effects of diencephalic 2 I. BRADY, J. V., AND D. CONRAD. Some effects of
stimulation of conscious animals in terms of normal limbic system self-stimulation upon conditioned
behavior. In: Neurological Basis of Behaviour, edited by emotional behavior. J. Camp. and Physiol. PsyclwZ. 53:
G. E. W. Wolstenholme and C. M. O’Connor. Lon- 128, 1960.
don: Churchill, I 958, p. 76. 22. BRECHER, R., AND E. BRECHER. The happiest
8. BEER, B., AND E. S. VALENSTEIN. Discrimination creatures on earth? Harper’s Magwine 222: 85, 1g6r.
of tones during reinforcing brain stimulation. Science 23. BROBECK, J. R. Mechanisms of the development
132: 297, 1960. of the obesity in animals with hypothalamic lesions.
g. BENNETT, I. F. Clinical studies with phenothiazene Physiol. Revs. 26 : 541, I 946.
derivatives in psychiatry. Research Pubis. Assoc. RG- 24. BRODIE, B. B., AND P. A. SHORE. A concept for a
search Nervous Mental Disease 37 : 266, I 959. role of serotonin and norepinephrine as chemical
IO. BICKFORD, R. G., D. DALY, AND H. M. KEITH. mediators in the brain. Ann. N. Y. Acad. Sci. 66: 631 J
Convulsive effects of light stimulation in children. * 957.
A.M.A. J. Diseases ChiZdren 86 : I 70, I 953. 25. BRODIE, D. A., J. L. MALIS, 0. M. MORENO,
II. BOWER, G. H., AND N. E. MILLER. Rewarding AND J. J. BOREN. Nonreversibility of the appetitive
and punishing effects from stimulating the same place characteristics of intracranial stimulation. Am. J.
in the rat’s brain. J. Camp. and Physiol. Psychol. 51 : Physiol. 199: 707, I 960.
6%’ 1958. 26. BRODIE, D. A., 0. M. MORENO, J. L. MALIS,

12. BRADY, J. V. Emotional behavior and the nervous AND J. J. BOREN. Rewarding properties of intra-
system. Trans. N. Y. Acad. Sci. I 8 : 7, 601, I 956. cranial stimulation. Science 131: 929, 1960.
I 3. BRADY, J. V. An experimental approach to the 27. BROOKHART J. M., AND F. L. DEY. Reduction
analysis of emotional behavior. In: Experimental of sexual behavior in male guinea pigs by hypothala-
Psychopathology, edited by P. Hoch and J. &bin. mic lesions. Am. J. Physiol. 133 : 551, 1941.
New York: Grune & Stratton’ Inc., 1957. 28. BROWN, G. W., AND B. D. COHEN. Avoidance
14. BRADY, J. V. The paleocortex and behavior moti- and approach learning motivated by stimulation of
vation. In: Biological and Biochemical Bases of Behavior, identical hypothalamic loci. Am. J. Physiol. I 97 : 153,
edited by H. Harlow and C. Woolscy. Madison: 1959.
University of Wisconsin Press, 1958. 29. BURE$ J. Reversible decortication and behavior.
15. BRADY, J. V. Temporal and emotional factors re- In : The Central Nervous System and Behavior, Transom-
lated to electrical self-stimulation of the limbic sys- tions of the Second Conference, edited by M. A. B.
tem. In: Reticular Formation of the Brain, edited by Brazier. New York: Josiah Macy, Jr., Foundation’
H. Jasper et al. Boston: Little, Brown & Co., 1958. 1959, Pa 207.
30. BURE& J., 0. BUREgOVA, E. FIFKOVA, J. thalamus and reticular formation. Acta Physiol. Acad.
OLDS, M. E. OLDS, AND R. P. TRAVIS. Spread- 6%. Hung. g : 133, 1956.
ing depression and subcortical drive centers. Physiol. 50. GREEN, J. D. A simple microelectrode for record-
Bohemosloven. I o : 32 I, I 961. ing from the central nervous system. NaZure I 82 : 962,
31* BURSTEN, B., AND J. M. R. DELGADO. Positive 1958.
reinforcement induced by intracranial stimulation in 51. GROSSMAN, S. P. Eating or drinking elicited by
the monkey. J. Camp. and Physz’ol. PsychoZ. 51: 6, 1958. direct adrenergic or cholinergic stimulation of hypo-
320 BUSH, R. R., AND F. MOSTELLER. Stochactic thalamus. S&ezce 132: 301, 1960.
Mod& for Lcaming. New York: John Wiley & Sons, 52* GRUNDFEST, H. Synaptic and ephaptic trans-
Inc., 1955. mission. In : Handbook of Physiology, edited by J. Field
330 COOK, L., AND E. WEIDLEY. Behavioral effects et al. Washington, D. C.: American Physiological
of some psychopharmacological agents. Ann. N. Y. Society, 1959, sec. I: Neurophysiology, vol. I, p. 147.
Acad. sci. 66: 740, 1957. 53- HEATH, R. G., AND W. A. MICKLE. Evaluation
340 CURTIS, D. R., AND K. KOIZUMI. Chemical of seven years’ experience with depth electrode
transmitter substances in brain stem of cat. J. Neuro- studies in human patients. In : EZectricaZ Studies on tb
physiol. 24: 80, 1961. Unanesthetized Brain, edited by E. R. Ramey and
35* DELGADO, J. M. R. Cerebral structures involved D. S. O’Doherty. New York: Paul B. Hoeber, Inc.,
in transmission and elaboration of noxious stimula- 1960, p. 214.
tion. J. Newophysiol. I 8 : 261, I 955. 54* HEBB, D. 0. Drives and the conceptual nervous
36. DELGADO, J. M. R., AND B. K. ANAND. In- system. Psychol. &v. 62 : 243, 1955.
crease of food intake by electrical stimulation of the 550 HESS, W. R. Diencephalon: Autonomic and Extra-
lateral hypothalamus. Am. J. Physiol. I 72 : 162, 1953. pyramidal Functions. New York : Grune & Stratton,
370 DELGADO, J. M. R., AND H. HAMLIN. Spontan- Inc., I 954.
eous and evoked seizures in animals and humans. 56. HESS, W. R. Das Zwischenhirn: Syndrome, LokaZisa-
In : EZectricaZ Studies on the Unanesthetized Brain, edited tionen, Functionen (2nd ed.). Basel: Schwabe, 1954.
by E. R. Ramey and D. S. O’Doherty. New York: 57* HESS, W. R., K. AKERT, AND D. A. MCDONALD.
Paul B. Hoeber, Inc., 1960, p. 133. The functions of the orbital gyri in cats. Brain 75 :
38. DELGADO, J. M. R., W. W. ROBERTS, AND N. 244,1952.
E. MILLER. Learning motivated by electrical atim- 58. HIGGINS, J. W., G. F. MAHL, J. M. R. DEE
ulation of the brain. Am. J. Physiol. I 79 : 587, I 934. GADO, AND H. HAMLIN. Behavioral changes dur-
390 DELGADO, J. M. R., H. E. ROSVOLD, AND E. ing intracerebral electrical stimulation. A.M. A.
LOONEY. Evoking conditioned fear by electrical Arch. NCUTOZ. P~y~iu’at. 76 : 399, I 956.
stimulation of subcortical structures in the monkey 59= HODOS, W., AND E. S. VALENSTEIN. Motiva-
brain. J. Camp. and Physiol. PsychoZ. 4g : 373, I 956. tional variables affecting the rate of behavior main-
400 DEMPSEY, E. W., AND D. McK. RIOCH. The tained by intracranial stimulation. J. Camp. and
localization in the brain stem of the oestrus responses Physiol. Psychol. 53: 502, 1960.
of the female guinea pig. J. Neurophysiol, 2 : g, 1939. 60. HODOS, W., AND E. S. VALENSTEIN. An evalu-
4** DEY, F. L., C. FISHER, 0. M. BERRY, AND ation of response rate as a measure of rewarding
S. W. RANSON. Disturbances in reproductive funcl intracranial stimulation. J. Camp. and Physiol. Psychd.
tions caused by hypothalamic lesions in female 55: 80, 1962.
guinea pigs. Am. J. PhysioZ. 129: 39, 1940. 61. HOEBEL, B. G., AND P. TEITELBAUM. Hype-
429 FERSTER, C. B., AND B. F. SKINNER. &h&r& thalamic control of feeding and self-stimulation.
of Reinforcemeti. New York : Appleton-Century- &iCnGG 135: 375, 1962.
Crofts, Inc., 1957. 62. HUBEL, D. H. Tungsten microelectrodes for re-
43* FISHER, A. Maternal and sexual behavior induced cording from single units. Science I 25 : 549, 1957.
by intracranial chemical stimulation. Scjencc 124: 63- HUGELIN, A., AND M. BONVALLET. Tonus
228, 1956. cortical et contile de la facilitation motrice d’origine
440 FONBERG, E., AND J. M. R. DELGADO. Inhibi- r&icuhire. J. PhysioZ. (London) 4g : I I 7 I, 1957.
tory effects of amygdala on food intake and condi- 64. HULL, C. L. Princ@Zes of Behavior. New York: Ap
tioning. (Abstr.) Federation Proc. 20: 335, I 961. pleton-Century-Crofts, Inc., r 943.
450 GASTAUT, H., R. NAQUET, R. VIGOUROUX, 65. HUNTER, J., AND H. H. JASPER. Effects of thal-
AND J. CORRIOL. Provocation de comportements mic stimulation in unanesthetized animals. He&o-
emotionnel s divers par stimulation rhinencephalique cncephalog. and Cl&z. Neuropiysiol. I : 305, 1949.
chez le chat avec electrodes a demeure. Rev. neuroZ. 66. JAHN, T. L., AND V. J. WULFF. Chemoreception.
86: 319, 1952. In : Comparative Animal Physiology, edited by C. L,
46. GAVLICEK, V. Electroencephalographic char- Prosser. Philadelphia: W. B. Saunders Co., 1950,
acteristics of conditioned defensive dominant state. P- 4479
Fi&oZ. Zhur. S. S. S. R. 4g : 4, I 958. 67* KAADA, B. R. Somato-motor, autonomic and elec-
trocorticographic responses to electrical stimulation
479 GLICKMAN, S. E. Reinforcing properties of arousal.
of rhinencephalic and other structures in primates,
J. Comb. and Physiol. Psychol. 53 : 68, 1960.
cat, and dog. Acta Physiol. Stand. 24: I, 1931.
48. GLOOR, P. Telencephalic influences upon the
68. KAADA, B. R., P. ANDERSEN, AND J. JANSEN,
hypothalamus. In : Hypothalamic HypophysiaZ ZnterreZa- JR. Stimulation of the amygdaloid nuclear complex
tionships, edited by W. S. Fields et al. Springfield, in unanesthetized cats. Neurology 4: 48, 1934.
Ill. : Charles C Thomas, 1956. KLING, J. UT., AND Y. MATSUMIYA. Relative
69.
499 GRASTYAN, E., K. LISSAK, AND F. KEKESI. reinforcing values of food and intracranial stimula-
Facilitation and inhibition of conditioned alimentary tion. Science I 35 : 668, I 962.
and defensive reflexes by stimulation of the hypo- 70. KLUEVER, H., AND P. C. BUCY. Preliminary
Analysis of Functions of the Temporal Lobes in 88. MACLEAN, P. D., AND D. PLOOG. Cerebral loci
Monkeys. A. M. A. Arch. Neural. Psychiat. 42: 979, involved in penile erection. (Abstr.) Federation Proc.
* 939. xg: 288, 1960.
71. KOPA, J., I. SZABO, AND E. GRASTYAN. A 89. MACLEAN, P. D., D. w. PLOOG, AND B. W.
dual behavioral effect from stimulating the same ROBINSON. Circulatory effects of limbic stimula-
thalamic point with identical stimulus parameters tion, with special reference to the male genital organ.
in different conditional reflex situations. Acta Physiol. Physiol. Revs. 40 (Suppl. 4) : 105, 1960.
Ad. Sci. Hung. In press. go. MACLEAN, P. D., B. W. ROBINSON, AND D. W.
72. LEA6, A. A. P. Spreading depression of activity in PLOOG. Experiments on localization of genital
the cerebral cortex. J. Neurophysid. 7 : 359, 1944. function in the brain. Trans. Am. Neurof. Assoc. 1959,
73. LILLY, J. C. Learning elicited by electric stimula- p. 105.
tion of subcortical regions in the unanesthetized gr . MAGOUN, H. W. Caudal and cephalic influences
monkey. Scz’ence I 25: 748, 1957. of brain stem reticular formation. Physiol. Revs. 30:
74. LILLY, J. C. Learning motivated by subcortical 459’ *950*
stimulation: the start and stop patterns of behavior. gn. MALMO, R. B. Slowing of heart rate after septal
In: Reticular Formation of the Brain, edited by H. self-stimulation in rats. ScieMe I 33 : I I 28, I 961.
Jasper et al. Boston: Little, Brown & Co., 1958, p. 93. MARGULES, D. L., AND OLDS, J. Identical “feed-
705. ing” and “rewarding” systems in the lateral hypo-
75. LILLY, J. C. Rewarding and punishing systems in thalamus of rats. Science 135: 374, 1962.
the brain. In: The Central Nervous System and Behavior, 94. MASSERMAN, J. H. Is the hypothalamus a center
Transactions of the First Conference, edited by M. A. B. of emotion? Psychosomat. Med. 3 : I, I 941.
Brazier. New York: Josiah Macy, Jr., Foundation, 95. MASSERMAN, J. H. The hypothalamus in psy-
‘959’ P. 247. chiatry. Am. J. Psychiat. 98: 633, 1942.
76. LILLY, J, C. Some considerations regarding basic 96. MILLER, N. E. Experiments on motivation. Scicnct
mechanisms of positive and negative types of motiva- 1271, 1957.
tions. Am. J. Psych&. I I 5: 498, I 958. 97. MILLER, N. E. Objective techniques for studying
77. LILLY, J. C. Stop and start effects. In: The Central motivational effects of drugs on animals. In: Psy-
Nervous System and Behaaior, Transactions of the Second chotropic Drugs, edited by S. Garattini and V. Ghetti.
Conference, edited by M. A. B. Brazier. New York: New York : Elsevier Pub. Co., 1957, p. 83.
Josiah Macy, Jr., Foundation, 1959, p. 76. 98. MILLER, N. E. Central stimulation and other new
78. LILLY, J. C. “Stop” and “start” systems. In: Neuro- approaches to motivation and reward. Am. Psy-
pharmacology, Transactions of the Fourth Conference, chologist I 3 : I 00, 1958.
edited by H. A. Abramson. New York: Josiah Macy gg. MILLER, N. E. Comments on the implications of
Jr., Foundation, 1959, p. 153. the Olds reward effect for theories of reinforcement.
79. LILLY, J. C. Learning motivated by subcortical In : EZectrical Stimulation of the Brain: Subcorticd Zn-
stimulation: the “start” and the “stop” patterns of tegrative Systems, edited by D. E. Sheer. Houston:
behavior. Injury and Excitation of Brain by Electrical University of Texas Press, 1961, p. 575.
Currents. In: Elcctricaf Studies on the &anesthetized I 00. MILLER, N. E. Integration of neurophysiological
Brain, edited by E. R. Ramey and D. S. O’Doherty. and behavioral research. In: Pavlovian Conference on
New York : Paul B. Hoeber, Inc., 1960, p. 78. Higher Nervous Activity. Ann. New York Acad. Sn’.
80. LILLY, J. C. The psychophysiological basis for two 92: 830, x96x.
kinds of instincts-implications for psychoanalytic IOI. MILLER, N. E., C. J. BAILEY, AND J. A. F.
theory, J. Am. Psychoanal. Assoc. 8 : I 54, I 960. STEVENSON. Decreased “hunger” but increased
81. LILLY, J, C., J. R. HUGHES, E. C. ALVORD, food intake resulting from hypothalamic lesions.
JR., AND T . W. GALKIN. A brief, noninjurious wave- Science I 12: 256, 1950.
form for stimulation of the brain. Science I 21: 468, 102. MORGANE, P. J. Distinct “feeding” and “hunger
1955. motivating” systems in the lateral hypothalamus of
82. LILLY, J. C., AND A. M. MILLER. Operant con- the rat. Science I 33 : 887, 1961.
ditioning of the bottlenose dolphin with electrical 103. MORRELL, F. Electrophysiological contributions
stimulation of the brain. J. Camp. and Physiol. Psychof. to the neural basis of learning. Physiol. Revs. 41:
55: 73, 1962. 443, 1961.
83. LINDSLEY, D. B. Emotion, In: Handbook of Ex- 104. NEWMAN, B. L. Behavioral effects of electrical
perimental Psychology, edited by S. S. Stevens. New self-stimulation of the septal area and related struc-
York: John Wiley 8z Sons, Inc., 1951, p. 473. tures in the rat. J. Camp. and Physid. Psychol. 54:
84. LORENTE DE NO, R. A study of nerve physiology. 340, *g6*.
Studies RockefelZer Inst. Med. Research 13 I and I 32 : 105. NIELSON, H. C., R. W. DOTY, AND L. T . RUT-
* 947. LEDGE. Motivational and perceptual aspects of
85. MACLEAN, P. D. Chemical and electrical stimula- subcortical stimulation in cats. Am. J. Physiol.
tion of hippocampus in unrestrained animals. Part 194: 427, *g58.
II. Behavioral findings. A. M. A. Arch Neural. Psy- 106. OLDS, ,I. Physiological mechanisms of reward. In:
chiat. 78: I 28, I 957. Nebraska Symposium on Motivation, ‘955, edited by
86. MACLEAN, P. D. The limbic system with respect to M. R. Jones. Lincoln: University of Nebraska Press,
self-preservation and the preservation of the species. 1955’ Pa 730
J. Nervous Mental Disease 127: I, 1958. 107. OLDS, J. A physiological study of reward. In:
87. MACLEAN, P. D., S. DUA, AND R. H. DENNIS- Studies in Motivation, edited by D. C. McClelland.
TON. Dien-mesencephalic loci involved in penile New York: Appleton-Century-Crofts, Inc., 1955,
erection and seminal discharge. (Abstr.) Federation P- 134
Proc. 20: 331, x96x. 108. OLDS, J. A preliminary mapping of electrical
reinforcing effects in the rat brain. J. Camp. and 127. OLDS, J,, AND M. E. OLDS. Positive reinforce-
Physiol. Psychot. 4g : 281, I 956. ment produced by stimulating hypothalamus with
fog. OLDS, J. Runway and maze behavior controlled iproniazid and other compounds. Science I 27 : I I 75,
by basomedial forebrain stimulation in the rat. ‘958.
J. Camp. and Physiol. Psychol. 49: 507, 1956. 128. OLDS, J., AND M. E. OLDS. Interference and
I IO. OLDS, J. Brain response to drugs mapped through learning in paleocortical systems, In: Bruin Mecha-
self-stimulation : In : Neuropharmacology, Transactions nisms and Learning, edited by J. F. Delafresnaye.
of the Third Confcre~cc, edited by H. A. Abramson. Oxford : Blackwell Scientific Publications, x961,
New York: Josiah Macy, Jr., Foundation, I 957, Pm 153.
P* 3430 129. OLDS, J., AND M. E. OLDS. Approach-avoidance
I I I. OLDS, J. Adaptive functions of paleocortical and interactions in rat brain. Am. J. Physiol. In press.
related structures. In : BiologicaZ and Biochemical 130. OLDS, J,, AND B. PERETZ. A motivational analysis
Bases of Behauior, edited by H. F. Harlow and C. N. of the reticular activating system. EZectroencephaZog.
Woolsey. Madison : University of Wisconsin Press, and CZin. NeurOphysioZ. 12: 445, 1960.
‘95% P* 237. 131. OLDS, J., AND J. C. SINCLAIR. Self-stimulation in
I I 2. OLDS, J. Effects of hunger and male sex hormone on the obstruction box. (Abstr.) Am. Psychologist 12 :
self-stimulation of the brain. J. Camp. and Physiol. 464, ‘957.

Psychol. 51: 320, 1958. 132. OLDS, J., AND R. P. TRAVIS. Relations of
I 13, OLDS, J. Satiation effects in self-stimulation of the carisoprodol and meprobamate to chlorpromazine
brain. J. Camp. and Physial. Psychol. 51: 675, 1958. in self-stimulation tests. In: The Pharmacology and
I 14. OLDS, J. Selective effects of drives and drugs on CZinicaZ Usefulness of Carisojwodol, edited by J. G.
“reward” systems of the brain. In: Neurological Basis Miller. Detroit: Wayne State University Press,
of Behauim, edited by G. E. W. Wolstenholme and ’ 959s P* 39*
C. M. O’Connor. Boston: Little, Brown & Co., 133. OLDS, J., AND R. P. TRAVIS. Effects of chlor-
*g58, P. 124. promazine, meprobamate, pentobarbital and mor-
I 15. OLDS, J. Self-stimulation experiments and dif- phine on self-stimulation. J. Pharmacol. ExptZ,
ferentiated reward systems. In : Reticular Formation Therap. I 28 : 397, rg6o.
of the Brain, edited by H. Jasper et al. Boston: Little, 134. OLDS, J., R. P. TRAVIS, AND R. C. SCHWING.
Brown & Co., 1958, p. 67 I. Topographic organization of hypothalamic self-
I 16. OLDS, J. Self-stimulation of the brain. &z&e I 27 : stimulation functions. J. Camp. and Physiol. Psychol.
315, ‘958. 53 : 23, ‘960.
I 17. OLDS, J. High functions of the nervous system. 135. OLDS, J., AND R. P. TRAVIS. Relative potency
Ann. Rev. PhysioZ. 2 I : 381, I 959. of phenothiazines by self-stimulation test (unpublished
I 18. OLDS, J. Self-injection in the rat brain. In: Neuro- observations).
psycho Pharmacology, edited by B. P. Bradley et al. 136. OLDS, J., A. H. YUWILER, M. E. OLDS, AND
New York: Elsevier Pub. Co., 1959. C. YUN. GABA microinjection in hypothalamic
I 19. OLDS, J. Studies of neuropharmacologicals by reward centers. (Abstr.) Federation Plot. 21 : 353,
electrical and chemical manipulation of the brain x96x.

in animals with chronically implanted electrodes. 137. OLDS, M. E., AND J. OLDS. Emotional and asso-
In: NeurOpsycho Pharmacology, edited by B. P. Bradley ciative mechanisms in rat brain. J. Camp. and Physiol.
et al. New York: Elsevier Pub. Co., 1959. Psychol. 54 : I 20, I 961.
120. OLDS, J. Approach - avoidance dissociations in 138. OLDS, M. E., AND J. OLDS. Approach-avoidance
rat brain. Am. J. Physiol. 199: 965, 1960. analysis of rat diencephalon. J. Camp. Neural. In
I 2 I. OLDS, J. Discussion of paper by C. W. Sem-Jacobsen press.
and Arne Torkildsen. In: EZectricaZ Studies on the 139. PAPEZ, J. W. A proposed mechanism of emotion.
Unanesthetized Brain, edited by E. R. Ramey and A.M.A. Arch. Neural. Psychiat. 38 : 725, x 937.
D. S. O’Doherty. New York: Paul B. Hoeber, Inc., 140. PAPEZ, J. W. Visceral brain, its component parts
1960, p. 290. and their connections. J. Nervous Mental Diseuse
122. OLDS, J. Operant conditioning of single unit I 26: 40, I 958.
responses. (Abstr.) Am. Psychologist I 5 : 480, I 960. 141. PORTER, R. W., D. G. CONRAD, AND J. V.
123. OLDS, J. Positive feedback mechanisms of the fore- BRADY. Some electroencephalographic patterns
brain. In: Symposium in Neural Mechanisms, Znforma- induced by self-stimulation in monkeys. Federation
tion Theory and Behavior, edited by M. H. Aprison et Proc. I 7: I 25, x958.
al. Battle Creek, Mich. : VA Hosp., March IO-I I, 142. PORTER, R. W., D. CONRAD, AND J. V. BRADY.
1960, p. 45 (mimeographed publ.). Some neural and behavioral correlates of electrical
124. OLDS, J., K. F. KILLAM, AND P. BACH-Y-RITA. self-stimulation in the limbic system. J. ExpiZ.
Self-stimulation of the brain used as a screening Analysis Behavior 2 : 43, 1959.
method for tranquilizing drugs. Science I 24: 265, 143. PROSSER, C. L. Nervous system. In: Comparative
‘956. Animal Physiology, edited by C. L. Prosser. Phila-
125. OLDS, J., K. F. KILLAM, AND S. EIDUSON. delphia: W. B. Saunders Co., 1950, p. 776.
Effects of tranquilizers on self-stimulation of the 144. REYNOLDS, R. W. The relationship between
brain. In: Psychotr~c Drugs, edited by S. Garattini stimulation voltage and rate of hypothalamic self-
and V. Ghetti. New York: Elsevier Pub. Co., 1957, stimulation in the rat. J. Camp. and Physiol. PsychOr.
P* 235. 51: ‘9% *g58-
126. OLDS, J., AND P. MILNER. P0Sh.h rehfOrCe- 145. ROBERTS, W. W. Rapid escape learning without
ment produced by electrical stimulation of septal avoidance learning motivated by hypothalamic
area and other regions of rat brain. J. Camp. and stimulation in cats. J. Camp. and Physiol. PsychoZ. 51:
Physiol. Psychol. 47: 419, 1954. 39’S ‘958.
14.6. ROBERTS, W. W. Both rewarding and punishing 161. STEIN, L., AND 0. S. RAY. Self-regulation of
effects from stimulation of posterior hypothalamus brain-stimulating current intensity in the rat.
of cat with same electrode at same intensity. J. Science 130: 570, 1959.
Camp. and Physiol. Psychol. 51: 400, I 958. 162. STEIN, L., AND 0. S. RAY. Brain stimulation re-
147. ROBERTS, W. W. Fear-like behavior elicited from ward “thresholds” self-determined in rat. Psy-
dorsomedial thalamus of cat. J. Camp. and Physiof. chopharmacologia I : 251, I 960.
Psychol. 55: 191, 1962. 163. STEIN, L., AND J. SEIFTER. Possible mode of
i48. SAWYER, C. H. Reproductive behavior. In: antidepressive action of imipramine. Stience 134:
Handbook of Physiology, edited by J. Field et al. Wash- 286, 1961.
ington, D. C.: American Physiological Society, I 64. TEITELBAUM, P. Sensory control of hypothalamic
1960, sect. I: Neurophysiology, vol. II, p. I 225. hyperphagia. J. Camp. and Physiol. Psychol. 48: 156,
149. SCHLOSBERG, H. Three dimensions of emotion. 1955.
Psychol. RGD. 61: 81, I 954. 165. TEITELBAUM, P., AND E. STELLAR. Recovery
IfjO. SEM-JACOBSEN, C. W., AND A. TORKILDSEN. from the failure to eat produced by hypothalamic
Depth recording and electrical stimulation in the lesions. Science x 20, 894, 1954.
human brain. In : Electrical Studies on the Unanesthetized 166. THOMPSON, R. The effect of damage to the in-
Brain, edited by E. R. Ramey and D. S. O’Doherty. terpeduncular nucleus on retention of a kinesthetic
New York: Paul B. Hoeber, Inc., I 960, p. 275. habit in rats. (Abstr.) Am. Psychologist 15: 482,
r51. SEWARD, J. P., A. UYEDA, AND J. OLDS. Re- 1960.
sistance to extinction following cranial self-stimula- 167. TINKLEPAUGH, 0. L. An experimental study of
tion. J. Camp. and Physiol. Psych& 52 : 294, 1959. representative factors in monkeys. J. Camp. Psych&
152. SEWARD, J. P., A. UYEDA, AND J. OLDS. Re- 8: 197, 1928.
*nforcing effect of brain stimulation on runway 168. TRAVIS, R. P., AND J. OLDS. rg5g. Two kinds of
performance as a function of interval between trials. escape from midbrain stimulation. (Abstr.) Am.
J. Camp. and Physiol. Psychol. 53 : 224, 1960. Psychologist I 4 : 430, I 959.
153. SHARPLESS, A. Designated discussion. In: Henry 169. WARD, H. P. Self-stimulation of tegmentum after
Ford HosfGtaI International Symfiosium, Reticular Forma- septal ablation. (Abstr.) Federation Proc. I 8: 166,
tion of the Brain, edited by H. Jasper. et al. Boston: 1959.
Little, Brown & Co., x958, p. 722. 170. WARD, H. P. Stimulus factors in septal self-stimula-
154. SIDMAN, M. Avoidance conditioning with brief tion. Am. J. Physiol. 196: 779, 1959.
shock and no exteroceptive warning signal. Science 171. WARD, H. P. Basal tegmental self-stimulation after
122: 157,rg53. septal ablation in rats. A.M.A. Arch. Neural. Psychid.
r55. SIDMAN, M., J. V. BRADY, J. BOREN, D. 3: 158, 1960.
CONRAD, AND A. SCHULMAN. Reward schedules 172. WARD, H. P. Tegmental self-stimulation after
and behavior maintained by intracranial self- amygdaloid ablation. A.M. A. Arch. Neural. Psychiat.
stimulation. Science I 22 : 3174, 830, 1955. 4: 657, 1961.
I 56. SKINNER, B. F. The Behauior of Organisms. New 173. WENDT, R., AND J. OLDS. Relations of drive
York : Appleton-Century-Crofts, Inc., 1938. and reward systems in the hypothalamus. (Abstr.)
1570 STEIN, L. Secondary reinforcement established with Federation Proc. 16: 136, 1957.
subcortical stimulation. Science I 27: 466, 1958. WILLIAMS, D. The structure of emotions reflected
174.
158. STEIN, L. Inhibitory effects of phenothiazene com- in epileptic experiences. Brain 79: 29, 1956.
pounds on self-stimulation of the brain. Diseases of
x75* WOLBARSHT, M. L., E. F. MAcNICHOL, JR.,
the NervousSystem. Suppl. 22: I, 1861.
AND H. G. WAGNER. Glass insulated platinum
r!i9* STEIN, L. Concurrent self-stimulation of two brain
microelectrodes. Science I 32 : 1309, 1960.
sites (unpublished manuscript).
I 76. WURTZ, R. H., AND J. OLDS. Chronic stimula-
160. STEIN, L., AND E. HEARST. Inhibitory effect of
positively reinforcing brain stimulation on learning. tion of amygdaloid complex. (Abstr.) Federation
(Abstr.) Am. Psychologist 13 : 408, 1958. Proc. 20 : 336, I 96x.

Hypothalamic Substrates of Reward

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hypothalamic Substrates of Reward

Uploaded by

Copyright:

Available Formats

Hypothalamic Substrates of Reward1

SYCHOLOGISTS have long identified “positive reinforcements” (i.e., rewards)

APPETITIVE BEHAVIOR PRODUCED BY ELECTRIC STIMULATION

In the rat, maximal effects are yielded by stimulation of a telencephalic-

FIG. I. Anatomical areas yielding negative reinforcement (dotted), positive reinforcement

A first step toward demonstrating that the neural tissuesstimulated in these

Operant Measures, “Self-Stimulation”

cycle/set were effective in yielding positive reinforcement with an optimum some-

Elicited Drive Responses

PUNISHMENT AND REWARD

The characteristic behaviors associated with the primitive property of irrita-

Anatomical Relations of Punishment and Reward Systems

Interaction of Motive Systems

Effects of CNS Damage

learned component of the escape behavior is sustained by the same mechanisms

Electrophysiological Ramzjications of the Stimulation

terior thalamic self-stimulation yielded similar waves with afterdischarges, and in

Operant Conditioning of Unit Responses

through the cortex, hippocampal formation, thalamus, and so forth, stopping

(sometimesseveral days) if the reinforcement procedure is used. Under no stimu-

ciated with positive reinforcement and sympathetic-like responses to be found in

states control reinforcement mechanisms is a common property of otherwise

stimulus in mid-maze caused an abrupt pause. The animal stopped in mid-

such an explanation certainly could not be applied to recent reports of impairment

E$ect of Drugs on Electric Stimulation

that a positive feedback processwas involved; a process,that is, which grows to a

Whether this selective action of chlorpromazine against the lateral hypo-

settings causedthe animal to distribute responsesevenly between the two levers. In

by injecting amphetamine. A similar selective action against negative mechanisms

The following chemicals were tested in solutions of I mg/ml of Ringer’s

by someprocessof neutralization so that stimulation ceasedand the animal came

would be unsuitable in peripheral ganglia or neuromyal junctions where tissues

stimulus constitutes a simple internal surrogate for either is unlikely. It seems

the reinforcing effect. An increment of excitation of these elements following

In summarizing such material as this, it does no harm to make a clear separa-

3) With electrodes in someplaces, the tendency to self-stimulate is a monotonic

Microelectrode studies indicate that paleocortical single-unit responseswhich

made of a neurophysiological response at a rewarding electrode site caused by

“Acquisition” is synonymous with “operant conditioning”; it is opposed to “extinction.”

6%’ 1958. 26. BRODIE, D. A., 0. M. MORENO, J. L. MALIS,

self-stimulation of the brain. J. Camp. and Physiol. 464, ‘957.

electrical and chemical manipulation of the brain x96x.

You might also like