Professional Documents
Culture Documents
1 (Spring)
place to turn. Smith goes on to say that liefs and desires. Changes in beliefs
the turn to cognitivism should be ac- and desires "can cancel, reverse, or
companied by a rejection of radical be- modify almost any environment-to-be-
haviorism (not to be confused with be- havior tendency in any number of
havior analysis). This is an issue to ways" (p. 95).
which I shall return. As Smith notes, others have also
Smith makes an important point made this argument. The important
when considering the criticism that the question for behavior analysts is
law of effect (better, the principle of whether it can be satisfactorily an-
reinforcement) is tautological. In the swered, and in chapter 6 Smith outlines
process of answering that criticism, he a rebuttal. It starts with folk psychol-
points out that the content of operant ogy's dependence on intentional expla-
psychology (behavior analysis) lies not nations that rely on such nonphysical
in the principle of reinforcement, but causes as beliefs and desires. Folk psy-
in the quantitative analysis of the prin- chology is then faced with the as-yet-
ciple of reinforcement: unsolved problem of the relation be-
tween physical and nonphysical onto-
although the property of functioning as a rein- logical spheres (see Baum & Heath,
forcer cannot be verified independently of the
occurrence of the response that is being rein- 1992, for more discussion of intention-
forced ... physical aspects of reinforcement al explanations). Furthermore, Smith
(e.g., the length of delay of delivery of rein- points out that the advocates of folk
forcement) can be verified and measured inde- psychology have misunderstood be-
pendently of certain physical aspects of the op- havior analysis, and when these mis-
erant response (e.g., the rate of operant respond-
ing). (p. 79) understandings have been removed, lit-
tle is left. Smith describes these mis-
It is these physical properties that are understandings and shows how, once
the independent and dependent vari- they are corrected, behavior analysis
ables of behavior analysis. remains unscathed. One misunder-
In chapter 5, Smith presents the folk standing has to do with the dependent
psychology critique of behavior anal- variable. Contrary to the critique, be-
ysis. He says that among philosophers havior analysts usually are not con-
this argument has been very influential. cerned with the occurrence of one re-
Basically, the argument is that even if sponse; rather, they are usually con-
every particular behavior (effect) has a cerned with patterns of responding
certain distal environmental cause, over time. Critics are also sometimes
there is no guarantee that a regularity unaware that the independent variable
connects the distal environmental also extends over time. Smith says that
cause to its behavioral effect. What is this is important, "because the tenden-
often required to be taken into account cy of folk psychology is to look at a
is an intervening process. Instead of particular response to a given stimulus
the distal-environmental-cause-to-dis- and note that if the organism were to
tal-behavioral effect, "we may get only change a certain belief, then the same
a regularity that says a certain distal stimulus would result in a different re-
[environmental] cause will produce a sponse" (p. 104). But behavior ana-
certain distal [behavioral] effect if the lysts predict behavioral patterns based
intervening process has certain prop- on the organism's exposure to the con-
erties" (p. 87). tingency (independent variable) for
Furthermore, in folk psychology, some duration of time.
many of these intervening processes Misunderstanding also stems from
are intentional in nature, comprising the failure of the advocates of folk psy-
persons' beliefs and desires. So in or- chology to distinguish between the
der to predict and control and under- functional categories behavior analysts
stand complex human behavior, we use to describe the environment/organ-
must ordinarily take into account be- ism system and the variables that are
ON BOOKS 197
the causes and effects connected by be- pose from that served by mentalistic
havioral principles. As Smith says, an categories.
operant response is not itself a depen- With respect to (c), the advantage
dent variable, and a reinforcing stimu- accruing to behavioral categories over
lus or discriminative stimulus is not it- mentalistic categories is that behavioral
self an independent variable. Rather, properties can be measured indepen-
these are the descriptive categories that dently of one another, but mentalistic
define the behavioral entities whose properties cannot. And in Smith's
measurable properties are the vari- view, "This independence of behavior-
ables. al quantities from one another is one
Some critics of behavior analysis, important benefit of doing psychology
according to Smith, also seem to be the behavioral way" (p. 123). It is an
completely unaware that functional important benefit because it enables
concepts play any role in behavioral behavior analysts to more easily estab-
generalizations. Dennett (1978) and lish quantitative relations experimen-
Taylor (1964), for example, have as- tally, and then to refine their under-
serted that behavioral generalizations standing of them.
can be refuted by the case in which an Smith also discusses (a) and (b),
organism misinterprets its environ- above. He analyzes examples that bol-
ment. Smith points out that this line of ster the suggestion that behavioral cat-
criticism completely ignores the role of egories are not subspecies of mental-
functional concepts in behavior analy- istic categories. The difference be-
sis. Functional concepts insure that the tween the categories is that mentalistic
organism is maintaining a certain type categories, involving beliefs and de-
of contact with the environment. But sires, have propositional content. For
when it is not in such contact, the prin- example, Bob believes that Neptune is
ciples do not apply. The criticism that the third planet from the sun and de-
behavior analysis fails to adequately sires to graduate from college. Behav-
account for the behavior of normal ioral categories, in contrast, describe
adult humans in complex everyday set- functional relations between the organ-
tings is also addressed. Here Smith ac- ism and the environment, and such de-
knowledges that behavioral principles scriptions do not have propositional
have limited predictive and control ca- content. Rather than saying that "The
pacity. This limitation should not, rat desires its lever presses to be fol-
however, be taken to imply that the lowed by pellets," the behavior analyst
principles are not valid in complex sit- says, "The rat's lever presses are re-
uations. inforced by access to pellets." Smith
In chapter 7, Smith maintains that shows that behavioral categories are
contemporary sophisticated philosoph- not logically connected to mentalistic
ical critiques of behavior analysis categories because behavioral concepts
question whether it is truly behavioral. are functional. There is no logical im-
The criticism is that behavior-analytic plication that if a certain mental state
and mentalistic concepts are somehow changes, then so does the function of
logically connected. Smith attempts to certain aspects of the behavioral sys-
distance behavior analysis from men- tem.
talism. There are (at least) three ways Part 3 is concerned with the
to do this: (a) Show that behavioral strengths and weaknesses of radical be-
categories are not a subspecies of men- haviorism. In chapter 8 ("What Is Rad-
talistic categories. (b) Show that not ical Behaviorism?"), Smith says that
only are behavioral categories not a radical behaviorism does not imply
subspecies of mental categories, they mind-body identity, although it elimi-
are not even logically connected with nates the mental, so radical behavior-
them. (c) Show that behavioral cate- ism is not dualistic. According to
gories serve a different scientific pur- Smith, the key to understanding Skin-
198 P. A. LAMAL
note in competition with those of folk develops), rather than as just another kind of
psychology and cognitive psychology. choice or preference task, there is a danger of
minimizing the importance of procedural details.
Smith, however, does not see it that There is much less danger of minimizing the im-
way. In his view, the recent growth of portance of things like training schedule, how
the behavior-analytic program has been preference is to be measured ... and what effect
limited, in contrast to cognitive psy- the particular training history might be expected
to have on the simple tendency to peck on the
chology, which has made rapid prog- ... different stimuli, if we regard the task as just
ress in accounting for complex learned another kind of choice procedure (p. 267).
behavior on the basis of underlying
cognitive processes. Behavior analysis, Higa and Staddon conclude their arti-
on the other hand, has not, indeed can- cle by again referring to the necessity
not, do this. of knowing the organism's relevant
Earlier research on schedules of re- history: "We should study the histori-
inforcement generated a collection of cal effects of training and carefully ex-
facts not easily related to one another. amine variables such as the presenta-
But with the matching law, operant tion order of stimulus pairs and rein-
psychology entered the realm of theo- forcement contingencies, as well as the
ry. "For the first time one can compare role of nonreinforced responses" (p.
with some precision the relative con- 291). Nothing cognitive here. Other
tribution of two or three different fac- terms in the JEAB articles that at first
tors to intentional behavior" (p. 205, glance may seem to refer to cognitive
emphasis added). In Smith's view, the processes do not do so when consid-
theory of operant psychology must be ered in the context of the articles in
cognitive, encompassing intentional id- which they appear.
ioms, and this is being recognized by In chapter 12 ("Operant Psychology
behavior analysts. He maintains, for Without Behaviorism"), Smith argues
example, that "Many behavioral psy- that for any given learned pattern of
chologists now concede that reference behavior, there are two quite different
to cognitive mechanisms is necessary types of explanation. One type of ex-
to provide explanations of behavioral planation provides a causal account of
regularities" (p. 215). But Smith's con- why the pattern arose. The other, re-
viction about this is stronger than the ductive, type of explanation argues that
evidence. For instance, I searched the the pattern is the result of a more basic
key words of all the empirical, theo- process. These are the transition and
retical, and review articles in eight is- property theories discussed above. So
sues of the Journal of the Experimental we seem to have a division of labor,
Analysis of Behavior (JEAB) published with operant psychology providing the
in 1993 and 1994. Of the 85 articles, first type of explanation and cognitive
only seven included one or more cog- psychology the second. Operant psy-
nitive terms in the listing of key words, chology and cognitive psychology are
hardly indicative of a floodtide of cog- thus complementary.
nitivism. Importantly, when a putative- They are also, however, in competi-
ly cognitive term (e.g., "remember- tion with one another. Both operant
ing") was used, it was described in a psychology and cognitive psychology
behavior-analytic, not cognitive, attempt to provide subsumptive expla-
framework. An important caveat is nations of everyday human behavior.
provided by Higa and Staddon (1993) Smith believes that operant psychology
in their study of "transitive inference" has two advantages in this competition,
in pigeons (note the quotation marks in which gives it a virtually insurmount-
the title of their article): able edge in accounting for ordinary
human behavior.
we argue that by considering the task [i.e., of One of operant psychology's advan-
transitive inference] as a measure of inference
based on the assumed ordering of stimuli (with- tages is epistemological. This derives
out providing a mechanism for how the ranking from the fact that behavioral units have
ON BOOKS 201
quantitative properties that can be mea- problem is not centered on the choice
sured directly. "The quantitative as- between naturalism (e.g., radical be-
pects of cognitive entities, on the other haviorism) and antinaturalism. "Rath-
hand, can be measured only indirectly er, it is to solve a series of puzzles
by making certain assumptions about about the implications of naturalism
the quantitative values of other psycho- for ethics. So it is odd to find Skinner
logical states" (p. 217). Such circular- asserting that since he has discovered
ity poses real difficulties for cognitive that behavior has natural causes, it fol-
psychology. lows that we should abandon the ideals
The other advantage accruing to op- of freedom and responsibility" (p.
erant psychology, according to Smith, 226). Smith also faults Skinner for fail-
is logical, and in his view, this is the ing to give serious consideration to
more important one. The advantage is compatibilism, the doctrine that free-
that behavioral concepts are capable of dom and responsibility are compatible
defining regularities that satisfy the with naturalism. Self-control and ethi-
principle of compositionality. This cal behavior can, Smith says, be ac-
means that "when more than one be- counted for and developed by behavior
havioral cause exerts influence over the analysis without any need to refer to
same variable, it is (at least theoreti- the tenets of radical behaviorism. This
cally) possible to calculate the net ef- part of the book is not clearly tied to
fect" (p. 218). The primary example of what has preceded it. Indeed, it could
this is the matching law. The law tells probably have been omitted.
us, for example, the degree to which a I take issue with Smith on two of his
delay of reinforcement for one re- fundamental theses: (a) that behavior
sponse will tend to shift responding to analysis must incorporate cognitive id-
alternative responses. The law also ioms, and (b) that radical behaviorism
tells us the degree to which an increase is independent from behavior analysis
in relative frequency of reinforcement (operant psychology) and that radical
for a response will have the opposite behaviorism impedes the development
effect. of behavior analysis and thus should be
The causal tendencies imparted by changes in
dispensed with. No persuasive evi-
delay and frequency of reinforcement maintain dence is presented that behavior ana-
their validity when they interact, and so they can lysts must adduce cognitive processes
be composed into a net result that tells us how in order to account for behavior, in-
much of an increase in frequency of reinforce- cluding complex human behavior. La-
ment will be necessary to offset a given increase beling such behavior with cognitive
in delay of reinforcement. (p. 218)
terms and appealing to unobserved,
It is not likely, however, that cognitive hypothesized processes have not been
principles can be composed this way. shown to advance cognitive psycholo-
Unless one has a complete description gy vis-'a-vis behavior analysis. There is
of a person's system of beliefs and de- also the question of just what are cog-
sires, prediction is unreliable and ex- nitive processes. Smith and other phi-
planation is incomplete. But a com- losophers of an intentional bent write
plete description of any person's sys- about beliefs and desires. Most cogni-
tem is probably impossible. tive psychologists, however, do not re-
In the last two sections of the book's fer to beliefs and intentions. It is not
last chapter, Smith briefly considers made clear by Smith just how beliefs
freedom and responsibility, self-con- and desires and other cognitive pro-
trol, and ethics. His thesis here is that cesses are to be conceptualized. It
radical behaviorism, unlike behavior seems to me that either they are phys-
analysis, has not been helpful. Contra iological or they are not. If they are
Skinner, radical behaviorism does not physiological, then their conceptualiza-
resolve the issues of free will and re- tion and explication should be in phys-
sponsibility because the philosophical iological terms. If they are not physi-
202 P. A. LAMAL