See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/8548352

Molar and Molecular Views of Choice

Article  in  Behavioural Processes · July 2004

DOI: 10.1016/j.beproc.2004.03.013 · Source: PubMed

CITATIONS READS
73 4,241

1 author:

William M Baum
University of California, Davis
143 PUBLICATIONS   6,599 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Reinforcement as Induction View project

Group choice and the Ideal Free Distribution View project

All content following this page was uploaded by William M Baum on 03 November 2017.

The user has requested enhancement of the downloaded file.

 Behavioural Processes 66 (2004) 349–359

Molar and molecular views of choice

William M. Baum∗
University of California, Davis 611, Mason #504, San Francisco, CA 94108, USA

Abstract
The molar and molecular views of behavior are not different theories or levels of analysis; they are different paradigms. The
molecular paradigm views behavior as composed of discrete units (responses) occurring at moments in time and strung together
in chains to make up complex performances. The discrete pieces are held together as a result of association by contiguity. The
molecular view has a long history both in early thought about reflexes and in associationism, and, although it was helpful to
getting a science of behavior started, it has outlived its usefulness. The molar view stems from a conviction that behavior is
continuous, as argued by John Dewey, Gestalt psychologists, Karl Lashley, and others. The molar paradigm views behavior as
inherently extended in time and composed of activities that have integrated parts. In the molar paradigm, activities vary in their
scale of organization—i.e., as to whether they are local or extended—and behavior may be controlled sometimes by short-term
relations and sometimes by long-term relations. Applied to choice, the molar paradigm rests on two simple principles: (a) all
behavior constitutes choice; and (b) all activities take time. Equivalence between choice and behavior occurs because every
situation contains more than one alternative activity. The principle that behavior takes time refers not simply to any notion of
response duration, but to the necessity that identifying one action or another requires a sample extended in time. The molecular
paradigm’s momentary responses are inferred from extended samples in retrospect. In this sense, momentary responses constitute
abstractions, whereas extended activities constitute concrete particulars. Explanations conceived within the molecular paradigm
invariably involve hypothetical constructs, because they require causes to be contiguous with responses. Explanations conceived
within the molar paradigm retain direct contact with observable variables.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Molar; Molecular; Paradigm; Choice; Local; Extended

1. Introduction are integrated wholes, existing not only in space but

Within behavior analysis exist two views of behav-
ior. The older view sees behavior as consisting of dis-
crete units, usually called responses. Since it treats
behavior as made up of bits and pieces this way, it
is aptly called molecular. The molar view of behavior
offers an alternative. It sees behavior as composed of
behavioral patterns (hereafter called activities) that, by
their very nature, are temporally extended. Activities
or patterns differ from discrete responses in that they
∗ Corresponding author. Tel.: +1-415-345-0050.
E-mail address: wmbaum@ucdavis.edu (W.M. Baum).
spanning time. Thus, instead of momentary responses
like the lever press or the footstep, the molar view sees
the activities of lever pressing and walking. Whereas
the molecular view names its discrete units with or-
dinary nouns, the molar view usually names activities
with gerunds.
2. The molecular view in historical perspective
The molecular view has a long history, because
much of its history is the history of associationism.
The notion of association of ideas goes back at least

0376-6357/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.beproc.2004.03.013
350 W.M. Baum / Behavioural Processes 66 (2004) 349–359
to Aristotle (Herrnstein and Boring, 1966). It was
developed significantly by John Locke (1632–1704),
who considered it not only the means to understand
the succession of ideas, but also the means to under-
stand the way that complex ideas are built up out of
simpler ones. Locke used expressions like “coalesce,”
“combination,” “connexion,” and “tying together” to
write about the association of simple ideas into com-
plex ones. James Mill (1773–1836) wrote more ex-
plicitly about the building up of complex from simple:
From a stone I have had, synchronically, the sensa-
tion of colour, the sensation of hardness, the sen-
sations of shape, and size, the sensation of weight.
When the idea of one of these sensations occurs, the
ideas of all of them occur. They exist in my mind
synchronically; and their synchronical existence is
called the idea of the stone; which, it is thus plain, is
not a single idea, but a number of ideas in a partic-
ular state of combination. (Herrnstein and Boring,
1966, p. 366)
Mill supposed that complex ideas also could con-
nect to make still more complex ideas; for example, he
wrote, “Brick is one complex idea, mortar is another
complex idea; these ideas, with ideas of position and
quantity, compose my idea of a wall” (Herrnstein and
Boring, 1966, p. 377).
Few of the associationists discussed action ex-
plicitly, the main exception being David Hartley
(1705–1757), who argued that the same principles of
association that applied to sensations and ideas applied
also to motions, and wrote about ideas and motions
more or less interchangeably, relying on a vibratory
notion of nervous activity to account for both. In sup-
posing that ideas and motions may be conjoined, he
anticipated Pavlov’s notion of the conditional reflex.
Association formed the backbone of scientific psy-
chology in the nineteenth century. The conception that
conscious experience could be analyzed into discrete
units that combined with one another to make up the
succession and complexity of experience seemed the
way to make sense of what otherwise seemed be-
yond comprehension. When Pavlov began studying re-
flexes, the concept of association must have occurred
to him inevitably as it did to Hartley. Behavior too
could be rendered comprehensible by analysis into dis-
crete units that combined with one another to make
up the succession and complexity of behavior. Since
the reflex was the only behavioral unit that was at
all well understood, behaviorists of the early twenti-
eth century incorporated it into their thinking about
behavior. Even before that, Thorndike saw the power
of the consequences of behavior to change its likeli-
hood to lie in their ability to strengthen the connec-
tion between stimuli and responses. He referred to
his experiments on the law of effect as “the exper-
imental study of associative processes” (Thorndike,
1911; Thorndike, 2000). Like Hartley and Mill before
him, he saw complexity as accretion of associations
and carefully distinguished between synchronicity and
succession:
We must not mistake for a complex association a
series of associations, where one sense-impression
leads to an act such as to present a new sense-
impression which leads to another act which in its
turn leads to a new sense-impression. Of the forma-
tion of such series animals are capable to a very high
degree. . . . By this power of acquiring a long series
animals find their way to distant feeding grounds
and back again. But all such cases are examples of
the number, not of the complexity, of animal associ-
ations. (Thorndike, 1911; Thorndike, 2000, p. 132)
In this quote, Thorndike explicitly anticipates the
notion of a chain of reflexes, written of by early be-
haviorists such as Hull. From such a notion, the step to
the behavioral chains of Keller and Schoenfeld (1950)
and Skinner (1953) was a small one.
Thus, the molecular view of behavior was an ad-
mirable start for a science. It handled two of the
most basic problems, complexity and succession, by
conceiving of behavior as composed of discrete units
(responses) that could be connected to one another
and to stimuli. The problem of complexity was ap-
proached by supposing that many simple units could
be connected together, and the problem of succession
was approached with the notion of chaining—that
one behavioral unit could be linked to a stimulus
produced by another behavioral unit. The molecular
view served in much the same way that corpuscular
mechanics served physics as it was getting started.
Eventually corpuscular mechanics outlived its useful-
ness, except as an approximation, and was replaced
with more continuous views like field theory, relativ-
ity, and quantum mechanics. Behavior analysis is at
a point in its development where it too should turn
 W.M. Baum / Behavioural Processes 66 (2004) 349–359
away from a corpuscular view of behavior, away from
discrete units toward more continuous concepts.
3. Criticisms of the molecular view
From the later 19th century, if not earlier, the
molecular view had its critics. John Dewey, for ex-
ample, wrote an article published in 1896 criticizing
the idea that the reflex could possibly serve as a
sound basis for understanding behavior. He argued
against the very idea that stimulus and response could
be conceived of as separate entities, maintaining in-
stead that behavior should be understood as entering
into coordinations that embrace both behavior and
environment. He wrote:
[T]he reflex arc idea, as commonly employed, is de-
fective in that it assumes sensory stimulus and mo-
tor response as distinct psychical existences, while
in reality they are always inside a coördination and
have their significance purely from the part played
in maintaining or reconstituting the coördination.
(Herrnstein and Boring, 1966, p. 323)
Arguing against viewing behavior as a “series of
jerks” and emphasizing instead the “unity of activity,”
Dewey wrote:
The ‘stimulus,’ the excitation of the nerve ending
and of the sensory nerve, the central change, are just
as much, or just as little, motion as the events taking
place in the motor nerve and the muscles. It is one
uninterrupted, continuous redistribution of mass in
motion. . . . It is redistribution pure and simple; as
much so as the burning of a log, or the falling of
a house or the movement of the wind. . . . There is
just a change in the system of tensions. (Herrnstein
and Boring, 1966, p. 324)
The coordination, he argued, is one continuous pro-
cess, “an organization of means with reference to a
comprehensive end.” This would be true of a “well
developed” instinct, such as a hen’s sitting on eggs,
or a “thoroughly formed” habit, such as walking. Of
these, he wrote:
There is simply a continuously ordered sequence
of acts, all adapted in themselves and in the order
of their sequence, to reach a certain objective end,
351
the reproduction of the species, the preservation of
life, locomotion to a certain place. The end has got
thoroughly organized into the means. In calling one
stimulus, another response we mean nothing more
than that such an orderly sequence of acts is taking
place. The same sort of statement might be made
equally well with reference to the succession of
changes in a plant, so far as these are considered
with reference to their adaptation to, say, producing
seed. It is equally applicable to the series of events
in the circulation of the blood, or the sequence of
acts occurring in a self-binding reaper. (Herrnstein
and Boring, 1966, p. 325)
Other writers repeated Dewey’s criticisms in var-
ious forms. The Gestalt psychologists Köhler (1947)
and Koffka (1935/1963) argued against the atomism of
associationism and urged the continuity of both expe-
rience and behavior. Holt (1965/1915), writing about
the Freudian wish, equated the wish with a “course
of action” and argued that “intelligent conduct, to say
nothing of conscious thought, can never be reduced to
reflex arcs and the like; just as a printing-press is not
merely wheels and rollers, and still less is it chunks of
iron” (p. 50). Those who seek to analyze behavior into
reflex arcs, he suggested, have “overlooked the form
of organization of these his reflex arcs, has left out
of the account that step which assembles wheels and
rollers into a printing-press, and that which organizes
reflex arcs” (p. 50). In a classic paper, “The prob-
lem of serial order in behavior,” Lashley (1961/1951)
criticized associative chain theories as inadequate to
account for the wholeness of extended sequences of
behavior. Although he drew primarily on examples
from verbal behavior, he pointed out:
Temporal integration is not found exclusively in
language; the coordination of leg movements in in-
sects, the song of birds, the control of trotting and
pacing in a gaited horse, the rat running the maze,
the architect designing a house, and the carpenter
sawing a board present a problem of sequences
of action which cannot be explained in terms of
successions of external stimuli. (p. 181)
He criticized the idea of associative chains by point-
ing out that in the enunciation of words a single sound
may be followed by a wide variety of sounds, depend-
ing on the word being spoken. Similarly, he wrote:
352 W.M. Baum / Behavioural Processes 66 (2004) 349–359
Words stand in relation to the sentence as letters do
to the word; the words themselves have no intrin-
sic temporal “valence.” The word “right,” for ex-
ample, is noun, adjective, adverb, and verb, and has
four spellings and at least ten meanings. In such a
sentence as “The mill-wright on my right thinks it
right that some conventional rite should symbolize
the right of every man to write as he pleases,” word
arrangement is obviously not due to any direct asso-
ciations of the word “right” itself with other words.
(p. 183)
Instead an utterance has a wholeness about it that
must arise from some sort of overall organization.
Lashley wrote:
There is a series of hierarchies of organization; the
order of vocal movements in pronouncing the word,
the order of words in the sentence, the order of
sentences in the paragraph, the rational order of
paragraphs in a discourse. Not only speech, but all
skilled acts seem to involve the same problems of
serial ordering, even down to the temporal coordi-
nation of muscular contractions in such a movement
as reaching and grasping. (p. 187)
All these criticisms of the molecular view, with
its concepts of building complexity by accretion of
discrete units and building temporal extension by the
linking of discrete units into chains, revolve around
two points. Firstly, they say that activities have a
continuity or wholeness about them that defies anal-
ysis into bits like the pieces of a puzzle. Neither is a
printing press a bunch of wheels and rollers, nor is a
sentence a bunch of words strung together word by
word. Secondly, they say that activities are organized
or coordinated with respect to ends or functions. One
cannot talk about activity coherently without naming
the function it serves. Behavior, they say, is continu-
ous, temporally extended, and organized.
One might think that we face an irresolvable di-
chotomy between two views of behavior, one atomistic
and one holistic. To an extent that is correct, but we
need not simply point to the inadequacies of atomism,
insist on holism, and let the matter drop there. A third
alternative exists that utilizes the ideas that behavior
is continuous, temporally extended, and organized,
but also allows analysis into parts without reverting
to atomism: the concept of levels of organization
embodied in the idea of nested temporally extended
activities.
4. The molar view as an alternative
In a similar discussion of entities of concern in ecol-
ogy, Buege (1997) sought to avoid both the atomism
of focusing exclusively on organisms and the holism
of seeing all nature as fundamentally indivisible. In-
stead, he suggested that the organism is one level of
organization, the species is another, and the ecosystem
is yet another, much as Lashley did with his “series
of hierarchies of organization.” To support his sug-
gestion, Buege distinguished between a class and an
individual, much as I did in an earlier paper (Baum,
2002). Buege points out that when people use the word
“class,” they often mean a collection of items. He de-
fines a collection as a grouping of individuals, the po-
tential value of which is equal to the sum of the values
of the members grouped. He gives as an example the
spare change to be found in someone’s pocket at any
particular time; its cash value is simply the sum of the
coins’ values. Skinner’s concept of the operant, which
he called a class, was probably a collection (Baum,
2002).
In contrast to a collection, Buege defines an
“individual” as an entity the potential value of which
exceeds the sum of the values of its parts. He gives as
an example a baseball team, which constitutes more
than just a collection of players. Individuals have at
least three properties that collections lack. Firstly,
each part stands in the relation of part to whole, as a
baseball player is part of a whole team. Secondly, the
parts cohere to function as a whole, as the team, not
the players, wins and loses games. Thirdly, individu-
als are subject to the attribute of scale, as a baseball
player is an individual at a different scale from the
team. Buege notes, “A being’s scale is the position
in which the being exists in relation to other things”
(p. 4). Thus, an organism, a species, and an ecosystem
are scaled in that order.
In an earlier paper I proposed that activities are
individuals (Baum, 2002). Activities satisfy Buege’s
three requirements. The parts of any activity are other
activities, less extended in time, that function together
so as to make the activity effective (to accomplish its
ends). Activities have the property of scale because
 W.M. Baum / Behavioural Processes 66 (2004) 349–359
the parts of an activity exist on a smaller scale than it
does, and the same activity as a part of another activity
exists on a smaller scale than that more extended one.
Swinging a tennis racquet, playing a point, playing a
game, playing a match, and preparing for the Olympics
are activities scaled in that order.
As Buege’s (1997) proposal that ecosystems are in-
dividuals with scale escapes the atomism/holism di-
chotomy, so too the molar view of behavior, seeing
it as composed of activities with scale, escapes the
atomism/holism dichotomy in behavior analysis.
A second feature of the molecular view of behav-
ior, besides its atomism, is its reliance on temporal
contiguity as the principle that binds the pieces or
discrete units together. Stimuli must be contiguous
with responses, and reinforcers must closely follow
responses to be effective. This requirement of stim-
uli and reinforcers immediately contiguous with re-
sponses leads the molecular view immediately into
positing hypothetical stimuli and reinforcers when it
comes to explaining behavior. For example, avoidance,
in which no consequences follow effective responses,
is explained by resorting to hypothetical reinforcers
in the form of fear reduction (e.g., Dinsmoor, 2001).
Rule-governed behavior, which depends on long-term
consequences, is explained by resorting to thoughts
and feelings (Malott, 2001; see Baum, 2003a, for crit-
icism).
The molar view offers an alternative to contiguity
for explanations of behavior. Instead, it suggests that
the environment provides correlations that manifest in
time, sometimes locally and sometimes on more ex-
tended scales. That a rat presses a lever at all may
be attributed to occasional episodes of pressing being
accompanied by episodes of feeding. A fixed-interval
pattern of pressing, however, results from exposure
to long-term regularity in presentation of reinforcers.
Saving money depends on an extended relation be-
tween activity and consequences, whereas spending
money on entertainment depends on more local rela-
tions entailed in a more local pattern (of socializing
perhaps).
Activities contrast with discrete responses in that,
rather than being combined like bricks, they com-
prise integrated parts. Every activity both entails parts
that are themselves activities and also is itself a part
in a more extended activity. Lever pressing is orga-
nized into bouts, which may go together to make
353
up a fixed-interval pattern, which is part of the rat’s
activities in the experimental chamber. Speaking a
sentence is part of conversing, which is part of main-
taining a relationship, which is part of living. In other
words, less extended activities are nested within more
extended activities.
5. Avoiding confusion
Activities are episodic; they occur in bouts or vis-
its. This might lead to confusion, because an episode
of an activity is a discrete event. Just like a discrete
response, it has a beginning and an end, and nothing
else happens in-between. Both may have variable du-
ration. So, what’s the difference? They differ in two
respects: (a) the way that duration enters into measure-
ment; and (b) the relation to reinforcers or rewards.
Firstly, duration is primary for activities, but sec-
ondary for discrete responses. Discrete responses vary
in duration, particularly if interresponse time (IRT) is
included. Researchers often have examined frequency
distributions of response and IRT duration. But each
response counts for just one response, albeit a re-
sponse with variable duration. The molecular view
treats the duration only as an attribute of the response.
In contrast, an episode of an activity counts according
to its duration—a brief episode adds less to the mea-
sure (time spent) than a long episode does. As an ex-
ample, in a choice experiment by Baum and Rachlin
(1969), in which no discrete responses were defined,
time spent on two platforms was measured by running
two clocks. Whenever the pigeon was on one side,
time on that side accumulated. A 5 s visit to a side
added 5 s, and a 30 s visit added 30 s. The episodes ac-
cumulated into time spent on the left and right sides.
Even if, as subsequent analysis suggests, choice con-
formed to a pattern of fix and sample (Baum, 2002),
time spent accumulated, except time spent fixing and
sampling. As we study IRT distributions, we also may
study frequency distributions of episode duration, but
the times involved in the two differ in their status,
one being a response attribute and the other being the
result of switching between activities.
Secondly, discrete responses are thought of as being
followed more or less immediately with a reinforcer
or reward. If reinforcement is continuous, every re-
sponse is followed by a reinforcer. If reinforcement is
354 W.M. Baum / Behavioural Processes 66 (2004) 349–359
intermittent, only some responses are followed by re-
inforcers, but each reinforcer is considered to follow a
response. Episodes of activity, in contrast, are thought
of as accompanied by reinforcers or rewards. A sin-
gle episode of pressing the left lever or pecking the
left key might include several reinforcers (Baum et al.,
1999). In the experiment by Baum and Rachlin, one
episode of standing on the left or rich side might in-
clude several reinforcers. In everyday life, this is true
in spades. An episode of sexual activity includes its
rewards, even if some additional ones may follow. The
same is true of an episode of reading.
Thus, a discrete response and an episode of an ac-
tivity differ in ontological terms. They have different
properties and play different roles in the conceptual
schemes or paradigms of which they are parts.
Distinguishing between the molar and molecular
views is complicated also by the long history of us-
age of the words molar and molecular. For example,
Hineline (2001) appears to consider the terms only to
anchor the ends of a continuum of scales of analysis,
probably referring to various scales of organization,
as we have discussed here. Confusion may be avoided
if these adjectives are used only to modify nouns
like view, thinking, and approach, rather than nouns
like analysis, control, or theory. When speaking of
analyses or control, one may instead speak of local
and extended analyses or short-term and long-term
control. Most likely, Hineline’s (2001) call for multi-
scaled analyses is a call for analyses at different scales
of temporal extendedness, from more local to more
global. Otherwise, his discussion seems in keeping
with the present discussion.
The reason to be careful applying the two adjec-
tives to theory is that the difference between the molar
and molecular views is not theoretical but paradig-
matic. The two approaches lead to different theories
and different methods, but they are themselves nei-
ther theories nor methods, because they are two dif-
ferent paradigms (Kuhn, 1970). A molecular theory
would mean a theory conceived in molecular terms.
For example, a molecularly conceived theory such
as Killeen’s (1994) behavioral mechanics competes
with other molecularly conceived theories such as
momentary maximizing (Hinson and Staddon, 1983)
in the sense that experiments may be designed that
pit them against one another. No such competition
occurs between these molecularly conceived theo-
ries, however, and molar-conceived theories such as
extended optimality (e.g., Baum, 1981; Rachlin and
Burkhard, 1978), because theories conceived in the
different paradigms are incommensurate—they are
unintelligible to one another, and no experiment could
decide between them. Theories conceived in different
paradigms differ both in form and in what they attempt
to explain. Instead of trying to explain the different
cumulative records generated by different schedules
of reinforcement, one may try to explain the allocation
of behavior between different choice alternatives.
In particular, although short-term relations may
exert more control over behavior than long-term re-
lations, such ascendancy in no way contradicts the
molar paradigm (Rachlin and Green, 1972; Reed
et al., 2003). Rather, the molar paradigm proposes the
question of short-term versus long-term control as a
question for study. What conditions shift control from
short-term to long-term and vice versa?
Although no experiment or data can decide between
the molar and molecular views, the two paradigms
may be tested in other ways. Either view may allow
an account of, say, avoidance (Herrnstein, 1969). One
will be favored over the other on the basis of plau-
sibility, elegance, and comprehensiveness. No exper-
iment can distinguish between the two-factor theory
of avoidance—a molecularly conceived theory—and
the shock-frequency-reduction theory—a molar-
conceived theory—but one may see the molar view of
avoidance as more plausible, more elegant, and more
comprehensive (Baum, 2001). The rest of this paper
presents arguments that aim to promote the molar
view on these grounds, focusing on choice.
6. Two simple principles
Two principles about behavior may be considered
fundamental: (a) all behavior constitutes choice; and
(b) all activities take time.
The idea that all behavior constitutes choice arises
from the recognition that every situation, no matter
how restricted, contains more than one behavioral
option (Baum, 1974). Even the most impoverished
experimental space, where only lever pressing or key
pecking is recorded, still permits the organism to en-
gage in other activities such as grooming, exploration,
and resting (Herrnstein, 1970). Choice is ubiquitous,
 W.M. Baum / Behavioural Processes 66 (2004) 349–359
because every situation offers multiple alternatives.
In a brief time frame, say a second or less, which
might have been taken up entirely by only one al-
ternative, other behavior could have occurred. In a
more extended time frame, other behavior will have
occurred, resulting in a mix of activities. In an exper-
iment on choice, at any particular moment, a pigeon
might peck at the left key or the right key or engage
in a number of other activities, but over the course
of minutes usually a mix of these different activities
occurs. In the activity of living, at any moment I
may be spending money or I may be saving money,
but over the course of a month I will do some of
both.
Recognition of the equivalence of behavior and
choice might seem to present no problem for the
molecular view. At any moment, one discrete response
or another occurs. Discrete responses translate into
discrete choices. If over time presses occur at the left
lever and the right lever, one may calculate the prob-
ability of a press on the left or the relative rate of left
presses. These calculations based on extended samples
of behavior, however, are regarded as “derived,” not
primary facets of behavior (Catania, 1981). The argu-
ment may be made that even response rate is a derived
measure, because one calculates it by picking a time
frame and then dividing the number of responses that
happen to occur by the duration (Dinsmoor, 2001). In
the molecular view, these measures, derived from the
concrete occurrences of discrete responses, are ab-
stractions. The lever presses are concrete particulars;
the relative rate of left presses—i.e., choice—is an
abstraction summarizing something about the presses.
In the molar view, behavior is choice and choice
is behavior. Whether local or extended, a relative
response rate constitutes an allocation of behavior
among alternatives and is a primary feature of behav-
ior. The measure may vary according to the duration
of the time frame and may vary from one measure-
ment to another with the same duration, particularly
when behavior is in transition from one situation to
another, but also even when behavior is stable. Never-
theless it estimates a feature of behavior that is funda-
mental. A coin comes up either heads or tails on any
one flip, for several flips different mixes of heads and
tails occur, and for a large number of flips the number
of heads may about equal the number of tails; the rel-
ative frequency of heads tells something fundamental
355
about the coin flipping. Allocations of behavior in a
situation, far from being derived abstractions, are the
primary data of the science.
The reason for these assertions in the molar view
lies in the second simple principle: that activities take
time. Even a lever press or key peck, so brief that the
molecular view seeks to treat it as having no duration,
still has duration, even if less than a second (Skinner,
1938). This is why measures of response rate never
come out infinite. In real life too even singular deci-
sions take time. When I decide to take a vacation in
Mexico, neither the decision nor the vacation is in-
stantaneous.
A deeper point, however, is that activities take up
time not just in practice but necessarily. Suppose I
show you a photograph of a person sitting and hold-
ing a book. What may we say the person is doing?
Is she reading? The correct answer is that you cannot
tell. She might be reading, but she might be pretend-
ing to read, or she might be daydreaming. You cannot
tell without more observation. You need to see what
went before and what comes after. If she is reading,
then afterwards she will be able to talk about the book.
I might reveal to you that she is actually pretending,
because before taking the picture I asked her to, so
she is complying with my request—a different activ-
ity altogether. To decide whether she is reading or do-
ing something else, you need more than a momentary
snapshot; you need a sample of behavior taking some
extended time. Indeed, for very extended activities,
you need a correspondingly lengthy sample. Is Billy
saving? If we just sample for a short time, we may see
him depositing money; a longer sample might show
us that, in fact, he was frittering his money away. To
quote Aristotle, “One swallow does not make a sum-
mer, nor does one day; and so too one day, or a short
time, does not make a man blessed and happy.” [The
sense of “make” here is that in the statement, “Three
interconnected lines make a triangle.”] (see Rachlin,
1994, p. 105).
The necessity of duration applies to any activity,
even to the preparations of the laboratory. If the pho-
tograph had contained a rat with its paw on a response
lever, and one asked what the rat was doing, the answer
would be the same. Does the rat remove its paw with-
out pressing? Did it press the lever just before? Unless
one sees what went before and what came after, one
cannot tell if the rat is pressing the lever, touching the
356 W.M. Baum / Behavioural Processes 66 (2004) 349–359
lever, or exploring. The point becomes even clearer if
we ask, when a press occurs, whether the press is part
of a fixed-interval pattern.
Thus, no sure identification of momentary behavior
is possible without an extended sample that brackets
the moment. As a result, the behavior of the moment
can be judged with certainty only in retrospect. Once
we have an adequate extended observation, we may
assert that at that moment the woman in the photo-
graph was complying with my request that she pretend
to read or that the rat was pressing the lever. Judg-
ments about momentary behavior necessarily take the
form, “At that moment, X was doing Y,” where Y is
always an extended activity that spans the moment.
The extended activity is the matter observed, the con-
crete particular. The momentary behavior is inferred,
an abstraction. Whereas the molecular view treats ex-
tended measures as abstractions and tries to treat mo-
mentary responses as concrete, the molar view takes
the extended activity as concrete and the momentary
response as an abstraction.
Significance attaches to this difference because of
the molecular view’s emphasis on contiguity. What-
ever behavior is occurring at the moment of occur-
rence of a reinforcer is supposed to be strengthened.
This requires that if no discrete response is apparent
one must be invented. Such invention leads to accounts
that are awkward and implausible when dealing with
continuous activities. Since a discrete response pre-
cedes the reinforcer, some chunk of reading must be
thought to precede, say, the garnering of a fact (if that
is a reinforcer). In contrast, an episode of an activity
comfortably brackets the reinforcer in a manner con-
sistent with the necessity of extension.
7. Nesting of activities
Activities are defined by their function. Walking
home is an activity, the function of which is to move
one through space toward one’s home. Let us say it
entails walking three blocks in one direction, turning
left, and walking five blocks again. Unless these parts
occur, the activity fails to perform its function. The
parts must be integrated and coordinated for the activ-
ity to work.
The parts of walking home, however, are themselves
activities, just less extended activities. Walking three
blocks along Market Street is an activity, the function
of which is to bring me to the corner at which I will
turn left. Walking five blocks to the left brings me the
rest of the way home. These activities in turn have
parts, such as crossing streets and waiting for traffic
lights, which must cohere in order to perform their
functions.
Thus, every activity both is composed of parts
which are less extended activities and is part of some
more extended activity. Walking home is part of ac-
complishing my tasks for the day. Reading has parts
such as turning pages and moving eyes. Reading
is a part of informing oneself. Pressing a lever has
parts such as moving a paw or contracting a mus-
cle. Pressing a lever is part of choosing between
two levers. The only exception would be an activity
such as living, which may be so extended as only
to have parts but to constitute no part of any more
extended activity, because no more extended activity
exists.
Fig. 1 shows a hypothetical illustration of nested
activities. Activities toward the left are more ex-
tended; those toward the right are more local. A
pigeon’s activities while in an experiment may be
broadly divided into three parts: feeding, maintaining
the body, and maintaining vigilance. Feeding is com-
posed of pecking and eating. Maintaining the body
is composed of resting and grooming. Maintaining
vigilance is composed of general activity and sensory
scanning. Each of these parts is composed of still less
extended activities. If pecking falls into a pattern of
fix and sample (Baum et al., 1999; Baum, 2002), then
it is composed of fixing on the rich alternative and
sampling the lean alternative.
Sampling on
Eating
Lean
Feeding Pecking Fixing on Rich
Experiment
Body Resting
Maintenance
Grooming
Vigilance General Activity
Sensory Scanning
Fig. 1. The hypothetical nesting of a pigeon’s activities in an
experiment with concurrent schedules of reinforcement.
 W.M. Baum / Behavioural Processes 66 (2004) 349–359
8. Molecular explanations: failings
The molar and molecular views lead to different
types of explanation. Why is the rat pressing the lever?
Why is the woman pretending to read? Any complete
answer to such a question must make reference to the
organism’s past—to the rat’s prior training and the
woman’s history with requests in general and my re-
quests in particular. The molecular view frames the an-
swers in terms of discrete stimuli, discrete responses,
and immediate reinforcers. The woman’s pretending
must be broken down into a sequence of pretending
responses, some of which must be followed by rein-
forcers. The molecular view must somehow bring the
past into the present by representing the past in stim-
uli immediately preceding and following responses.
The past, however, is invisible. The reinforcers for the
woman’s compliance are invisible; even my thanks at
the end—if they constitute a reinforcer—are too de-
layed to explain the entire episode, which lasted sev-
eral minutes. Thus, as with avoidance, the molecular
view resorts to hypothetical stimuli and reinforcers to
treat the compliance as a chain of discrete stimuli and
responses held together by hypothetical reinforcers.
In the laboratory, a chain may be arranged by causing
responding to produce a discriminative stimulus for
some other responding, which produces a reinforcer.
The molecular view takes the discriminative stimulus
produced to be a conditional reinforcer (analogous
to a Pavlovian conditional stimulus), and explains a
real-world episode of compliance by imagining that
it is a chain similarly held together by discrimina-
tive stimuli that double as conditional reinforcers
(see Baum, 1973, for discussion). Relying solely on
the concepts of discrete responses and contiguity of
reinforcers forces such awkward results.
9. Molar explanations: advantages
Both views offer ways of accounting for behav-
ioral episodes like the woman’s compliance, but, in
keeping with the paradigmatic nature of the differ-
ence between them, their accounts differ in elegance,
plausibility, and comprehensiveness. As the molar
view allows explanation of avoidance by citing the
temporally extended dependence of shock rate on
response rate (Herrnstein, 1969), so too it allows
357
explanation of the episode of pretending to read by
citing the woman’s long history of complying with
requests of people with whom she interacts socially,
her shorter history of interacting with me socially,
and her complying with my request still more locally
(Baum, 1995, 2003b). The problem of bringing the
past into the present never arises, because her activ-
ities past, present, and future are seen as continuous.
Her socializing contains as a part her interacting with
me, which includes as a part her complying with my
request. The reinforcers for socializing have been and
continue to be many and varied, accompanying the
various parts of socializing with different people in
her acquaintance. They take the form both of thanks
and reciprocation. Next time she asks me for a favor, I
will be likely to deliver consequences that, if remote in
time, nevertheless directly stem from her compliance
on that occasion. Our complying with one another’s
requests may be parts of our knowing one another.
Thus, in the molar view, less extended (i.e., more
local) activities are explained by their fitting as parts
into more extended activities (Rachlin, 1994). For
example, the various parts of maintaining a social
relationship are less extended activities that may pro-
duce their various rewards but may also enable more
extended activities in which they participate to pro-
duce otherwise unattainable rewards. A conversation
may be highly reinforcing, and receiving help from
a friend is also reinforcing. The reinforcing value of
the whole, however, may exceed the sum of the parts’
reinforcers, because some parts may produce rewards
that depend on the occurrence of other parts. Helping
a friend may produce reinforcers (thanks and obliga-
tion) that depend on the likelihood that the friend will
help on another occasion in return. Indeed, some parts
may produce no reward or even be aversive but still
may be essential to the value of the extended activity.
Turning down a piece of cake on any particular occa-
sion, a difficult and onerous activity, may be essential
to maintaining a diet that has huge reinforcing value
in the long run (Rachlin, 1995). In contrast with the
molecular view, this approach to explanation makes
no use of hypothetical events or contingencies. The
molar view refers to concrete, demonstrable stimuli
and consequences that, however, are usually visible
only over extended periods of time.
The molecular view resorts to hypothetical causes
also when it attempts to explain the occurrence of
358 W.M. Baum / Behavioural Processes 66 (2004) 349–359
particular responses on particular occasions. If a lever
press is preceded by no observable stimulus, the
molecular view explains it as the result of response
strength or arousal, hypothetical causes (Skinner,
1938; Killeen, 1994). The molar view, in contrast,
requires no account of momentary occurrences, be-
cause living, with all its parts, is continuous. Par-
ticular episodes of an activity, such as pressing or
complying with a request, are occasioned by changes
in the environment, such as the passage of time or
the occurrence of the request. Nothing analogous to
response strength comes in, even when an activity
undergoes extinction. When lever pressing no longer
produces food, an allocation including substantial
lever pressing transforms into one that includes little
or none, as other activities increase in frequency. The
molar view requires no notion like strength, because
in the molar view reinforcement shifts the allocation
of activities directly. Competition occurs, not between
different types of discrete responses, but between dif-
ferent allocations of activities (Baum, 2002). If one
allocation produces more reinforcers in the long run,
that allocation will tend to dominate in the long run
(Baum, 1981).
10. Conclusions
By restricting the terms molar and molecular to refer
only to the two different paradigms and by referring
to analyses or relations that differ in their temporal
scale as varying in extendedness, we gain the ability to
contrast the two views without confusing theoretical
or methodological considerations with paradigmatic
considerations. Although the two paradigms lead to
different phenomena and theories, no one should take
them to be theories or methods themselves. Recog-
nizing two simple principles, that all behavior entails
choice and that all activities take time, illuminates ad-
vantages of the molar view. The first advantage is that
these principles point up the possibility of conceiving
of allocation of time among activities as the funda-
mental measure of behavior. That idea, in turn, affords
a natural way of talking about behavior both in the lab-
oratory and in the real world, without resorting to im-
plausible discrete response units. It offers instead the
understanding and explaining of activities as parts of
wholes that are more extended activities—the concept
of nesting. Finally, it allows explanations that depend
on no hypothetical events but only on observable and
measurable aspects of behavior and environment.
Acknowledgements
The author thanks Michael Davison for many help-
ful comments on earlier versions.
References
Baum, W.M., 1973. The correlation-based law of effect. J.
Exp. Anal. Behav. 20, 137–153.
Baum, W.M., 1974. Choice in free-ranging wild pigeons. Science
185, 78–79.
Baum, W.M., 1981. Optimization and the matching law as accounts
of instrumental behavior. J. Exp. Anal. Behav. 36, 387–403.
Baum, W.M., 1995. Rules, culture, and fitness. Behav. Anal. 18,
1–21.
Baum, W.M., 2001. Molar versus molecular as a paradigm clash.
J. Exp. Anal. Behav. 75, 338–341.
Baum, W.M., 2002. From molecular to molar: a paradigm shift in
behavior analysis. J. Exp. Anal. Behav. 78, 95–116.
Baum, W.M., 2003a. The molar view of behavior and its usefulness
in behavior analysis. Behav. Anal. Tod. 4, 78–81.
Baum, W.M., 2003b. Understanding Behaviorism: Science,
Behavior, and Culture. Blackwell Publishing, Oxford (originally
published in 1994).
Baum, W.M., Rachlin, H.C., 1969. Choice as time allocation. J.
Exp. Anal. Behav. 12, 861–874.
Baum, W.M., Schwendiman, J.W., Bell, K.E., 1999. Choice,
contingency discrimination, and foraging theory. J. Exp. Anal.
Behav. 71, 355–373.
Buege, D.J., 1997. An ecologically-informed oltology for
environmental ethics. Biol. Phil. 12, 1–20.
Catania, A.C., 1981. The flight from experimental analysis. In:
Bradshaw, C.M., Szabadi, E., Lowe, C.F. (Eds.), Quantification
of Steady-State Operant Behavior. Elsevier, North-Holland,
Amsterdam, pp. 49–64.
Dinsmoor, J.A., 2001. Stimuli inevitably generated by behavior that
avoids electric shock are inherently reinforcing. J. Exp. Anal.
Behav. 75, 311–333.
Herrnstein, R.J., 1969. Method and theory in the study of
avoidance. Psychol. Rev. 76, 49–69.
Herrnstein, R.J., 1970. On the law of effect. J. Exp. Anal. Behav.
13, 243–266.
Herrnstein, R.J., Boring, E.G., 1966. A Source Book in the History
of Psychology. Harvard University Press, Cambridge, MA.
Hinson, J.M., Staddon, J.E.R., 1983. Matching, maximizing, and
hill-climbing. J. Exp. Anal. Behav. 40, 321–331.
Hineline, P.N., 2001. Beyond the molar-molecular distinction: we
need multiscaled analyses. J. Exp. Anal. Behav. 75, 342–347.
 W.M. Baum / Behavioural Processes 66 (2004) 349–359
Holt, E.B., 1965. The Freudian Wish and Its Place in Ethics.
Johnson Reprint Corp., New York (originally published in 1915
by Henry Holt and Co., New York).
Keller, F.S., Schoenfeld, W.N., 1950. Principles of Psychology.
Appleton-Century-Crofts, New York.
Killeen, P.R., 1994. Mathematical principles of reinforcement.
Behav. Brain Sci. 17, 105–171.
Koffka, K., 1935. Principles of Gestalt Psychology. Harcourt,
Brace, & World, New York (reprinted in 1963).
Köhler, W., 1947. Gestalt Psychology. The New American Library,
New York.
Kuhn, T.S., 1970. The Structure of Scientific Revolutions, second
ed. University of Chicago Press, Chicago.
Lashley, K.S., 1961. The problem of serial order in behavior. In:
Saporta, S. (Ed.), Psycholinguistics: A Book of Readings. Holt,
Rinehart, and Winston, New York, pp. 180–198 (originally
published in 1951).
Malott, R.W., 2001. Moral and legal control. Behav. Dev. Bull. 1,
1–7.
View publication stats
359
Rachlin, H., 1994. Behavior and Mind: The Roots of Modern
Psychology. Oxford University Press, New York.
Rachlin, H., 1995. Self-control: beyond commitment. Behav. Brain
Sci. 18, 109–159.
Rachlin, H., Burkhard, B., 1978. The temporal triangle: response
substitution in instrumental conditioning. Psychol. Rev. 85, 22–
47.
Rachlin, H., Green, L., 1972. Commitment, choice, and self-
control. J. Exp. Anal. Behav. 17, 15–22.
Reed, P., Hildebrandt, T., DeJongh, J., Soh, M., 2003. Rats’
performance on variable-interval schedules with a linear
feedback loop between response rate and reinforcement rate. J.
Exp. Anal. Behav. 79, 157–173.
Skinner, B.F., 1938. Behavior of Organisms. Appleton-Century-
Crofts, New York.
Skinner, B.F., 1953. Science and Human Behavior. Macmillan,
New York.
Thorndike, E.L., 2000. Animal Intelligence. Hafner Publishing Co.
(originally published in 1911 by Macmillan, New York).