JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2004, 82, 73–78 NUMBER 1 ( JULY)

THE ACCIDENTAL BEHAVIORIST: A REVIEW OF THE NEW BEHAVIORISM BY

Staddon contends that behavior analysis may profit from theory, by which he means theory that
includes assumed elemental processes allowing derivation of observed patterns of results. Behavior
analysts who consider that our field has matured to the point where we are secure in our descriptive
grasp of a range of phenomena will tend to agree. Unfortunately, the book’s positive message is
almost lost in a morass of distracting criticisms of Skinner and behavior analysis in general. Instead
of recognizing the growth and maturation of the field, Staddon writes as if the field has stagnated
and blames that stagnation on what he takes to be Skinner’s antipathy toward theory. Neither has
behavior analysis become stagnated, nor should Skinner be blamed for any shortcomings. Instead
of acknowledging Skinner’s foundational contributions to the field, however, Staddon devotes most
of this book to bashing Skinner and fails to distinguish his own view of theory based on internal
states from theory in cognitive psychology, treating theoretical models and unseen processes as the
‘‘true’’ subject of inquiry and behavior as only an indicator.
Key words: Skinner, behaviorism, theory, internal states, Staddon

This book is about theory. By structure and ory of Mind, argued that, from a logical point
substance, it makes an argument in favor of of view, intentional terms such as know, want,
theory in behavior analysis, a view that has and believe are not causative but summative.
been controversial for over 50 years. Whether Part of Tom’s belief in the uptown bus is that
one agrees with the argument or not, the he takes it. Part of his wanting to go home is
book raises old questions: What is a theory, that he goes home. Another part of his belief
and what makes a satisfactory theory? would be Tom’s talking about the uptown bus
Probably every science began with some as a means to get home. Instead of his taking
sort of folk version. No doubt behavior anal- the bus and talking about that action consti-
ysis emerged from folk psychology. The key tuting evidence of his inner belief, both the
terms of folk psychology are what philoso- action and its description are his belief. Ryle
phers call intentional: terms such as ‘‘know,’’ criticized the idea that inner belief causes be-
‘‘want,’’ and ‘‘believe’’ that seem to imply an havior as an example of what he called the
agent who knows, wants, and believes. If we paramechanical hypothesis: the idea that
say that Tom takes the uptown bus because mind mechanically causes behavior. He
he wants to go home and believes that bus traced the paramechanical hypothesis to the
will take him there, then it is an explanation notion that the inner mind animates the ex-
of sorts. One might say it is based on a theory: ternal body, which he mockingly referred to
that when an agent wants something and be- as the notion of the ‘‘ghost in the machine.’’
lieves a course of action will achieve that goal, Skinner (1950/1968) also famously at-
then the agent will adopt that course of ac- tacked the making of theories that invoke in-
tion. Why do we object to such a theory? ner causes to explain behavior. In an oft-quot-
The reason is easy to come by. It assigns ed passage, he set out to criticize theories of
desire and belief the status of causes, and we a certain sort:
have no idea how a desire to go home or a
. . . any explanation of an observed fact which
belief in a bus could cause Tom to take the
appeals to events taking place somewhere else,
bus. Gilbert Ryle (1949), in his book The The- at some other level of observation, described
in different terms, and measured, if at all, in dif-
The new behaviorism: Mind, mechanism, and society, by ferent dimensions. (Skinner, 1950/1968, p. 4)
John Staddon. Published in 2001 by Taylor and Francis:
211 pages, with references, author index, and subject in- With this definition, he rejected theories
dex.
Correspondence should be addressed to William M. appealing both to mental causes and also to
Baum, 611 Mason #504, San Francisco, California 94108 observed or supposed events in the nervous
(e-mail: [email protected]). system—in other words, theories that were

73
74 WILLIAM M. BAUM

‘‘neural, mental, or conceptual’’ (p. 5). From Oddly enough, Skinner (1986) objected to
the context, one gathers that Skinner object- the matching law. Perhaps the reason was that
ed to theories that depend on events in the he actually had a theory, but of a different
mind or nervous system (‘‘somewhere else’’), sort than a formal representation of data. He
observed by introspection or electrodes in had an idea of a mechanism that would gen-
the brain (‘‘at some other level’’), described erate all those beautiful cumulative records.
in terms of ideas, processes, or synapses (‘‘dif- In his famous paper on ‘‘Superstition in the
ferent terms’’), and either never measured Pigeon,’’ and in the book with Ferster, he
(mental causes) or measured in dimensions adopted the theory that delivery of a rein-
having nothing to do with behavior (e.g., forcer reinforces whatever behavior immedi-
membrane potentials). ately precedes it and proposed that this
In historical perspective, sciences appear to mechanism would account for all changes in
adopt different standards about theories at behavior (Ferster & Skinner, 1957; Skinner,
different stages in their development. Skin- 1948/1968). Researchers who study the
ner’s critique of learning theories arose be- matching law are similarly dissatisfied with
cause he saw that traditional theories about only formal representation; they want to
behavior had outlived their usefulness in the know what underlying mechanism might ac-
light of new methods of generating data and count for the observed regularity.
of new types of data being generated. The Hence we may infer that another type of
possibilities for studying operant behavior re- theory exists besides formal representation:
corded with counters and cumulative records theories that allow one to derive the formal
seemed virtually endless. They totally eclipsed representation. Although one could argue
the old runways and mazes. With all these that such a theory is itself a formal represen-
new data, Skinner was saying, we have no tation, for our discussion of what a theory is
need of the old scaffolding, because we can and what is a satisfactory theory, we may use-
develop new theories that consist of ‘‘a for- fully distinguish between a purely descriptive
mal representation of the data reduced to a summary and a theory that allows one to de-
minimal number of terms’’ (p. 21). As he rive that descriptive summary from more el-
never actually offered such a theory himself, emental processes. Such a theory still may es-
his suggestion remains open to interpreta- cape Skinner’s strictures, because the
tion, and 50-odd years later, we may wonder elemental processes (reinforcer-response
whether we have developed such theories and contiguity for Skinner, and, for example, re-
whether we might be ready for some other inforcer tracking for the matching law [Dav-
types. ison & Baum, 2000]) may take place in, and
No doubt operant methods have generated be measured in terms of, behavior.
a great deal of data. One may easily set up So, what about Staddon’s view—theoretical
experiments and drown in data. Since Skin- behaviorism? It amounts to the idea that the-
ner, perhaps as a defense against drowning, ories of the elemental sort (that allow deri-
we have developed formal representations vation of data) are permissible in behavior
that help us to ask more pointed questions. analysis. In particular, Staddon advocates dy-
The book Schedules of Reinforcement, by Ferster namic, real-time theories. He presents as an
and Skinner (1957), illustrates a certain type example a model of habituation employing
of exploratory research, based on asking the the notion of cascaded leaky integrators that
question, ‘‘What if I do this?’’ over and over. he has presented a few times before. It is a
It contains an enormous quantity of infor- good example because it allows prediction of
mation, but hardly organized at all, except by a seemingly paradoxical result: that even
the schedules tried. Nowadays, we have a though high rates of stimulation produce
more quantitative analysis, stemming mostly greater habituation, the habituation pro-
from Herrnstein’s (1961) discovery of the duced by lower rates of stimulation dissipates
matching law. Indeed, the matching law, in more slowly. I can easily imagine the satisfac-
its various forms (Baum, 1974, 1979; Herrn- tion the accomplishment must have brought.
stein, 1970, 1974), would seem to be exactly To read Staddon’s book, however, one
the sort of formal representation of data that would think that no other behavior analyst
Skinner advocated. looks with favor on such theories. He refers
 A REVIEW OF THE NEW BEHAVIORISM 75

to almost no other behavior analyst except in the way our model says it does’’ (Staddon,
himself and his students. He writes as if he 2001a, p. 153). Despite his claims to the con-
alone sees how to bring theory to the science trary, Staddon, like the cognitivists, thinks he
and as if all behaviorists since Hull and Tol- is studying states and mechanisms in the
man have labored in the darkness of an ir- brain.
rational and implacable hostility to theory. As the standard for deciding among mod-
Here is Staddon to save the day. But never els, Staddon shies away from prediction, and
mind; the important question is how well he appeals instead to parsimony. Parsimony,
makes the case. however, like aesthetics, has its own problems;
The weaknesses of the book stem from one person’s parsimony is another person’s
Staddon’s structuring it in light of ‘‘old oversimplification. Perhaps aware of this,
wrong, new right.’’ He overlooks the devel- Staddon waffles and winds up equating mod-
opment of the field. Research in behavior els with formal representations of data, the
analysis has grown quantitative, particularly in view of which Skinner, inspired by Mach’s
areas such as choice, detection, timing, and (1883/1960) Science of Mechanics, also ap-
behavioral economics. These developments proved. Staddon comments disingenuously,
have taken us to the point where theory ‘‘a formulation not far from TB’’ (theoretical
makes sense—is no longer a distraction, a behaviorism; Staddon, 2001a, p. 154). More
snare, and a delusion. Instead of considering like indistinguishable—but that would raise
that Skinner’s strictures about theory may the question of what all the fuss is about.
have been constructive in 1950 and that the As I was reading the book, that question
field has developed since to the point where occurred to me many times. The most bizarre
it is ready for theory, Staddon takes to bash- and troubling aspect of it is that, following an
ing Skinner, lending the book a sadly nega- initial chapter praising Hull and Tolman, ful-
tive tone. For example, the quotation about ly four chapters—half the book—are devoted
‘‘events taking place somewhere else,’’ like to assessing and mostly attacking Skinner. In-
most of Skinner’s writing is open to multiple stead of presenting his own views in a positive
interpretations, but Staddon chooses to inter- way, Staddon goes on page after page exco-
pret it in only one way: as forbidding virtually riating Skinner. I found myself objecting,
any type of theory. In historical context, Skin- even though I agree with most of the criti-
ner’s statement may be seen to refer to pre- cisms. I agree that Skinner indulged in pre-
mature modeling in the absence of data, the mature extrapolation, that he was wrong
sort of theories that learning psychologists about the facts on punishment, that he made
were spinning with scant support from data. unlikely, possibly irresponsible, recommen-
Staddon, however, uses his special interpre- dations. Staddon’s critique of Skinner’s
tation to call Skinner’s warning ‘‘antitheoret- (1948/1968) superstition paper as a piece
ical,’’ ‘‘bizarre,’’ and ‘‘theoretical seppuku.’’ more polemical than scientific is excellent,
The best chapters are the last three: one except for the tone. There and throughout,
on cognitive psychology, one on theory, and his choice of words is sarcastic and even
one on consciousness. Any behaviorist will en- snide. When he calls Skinner a rhetorical ge-
joy the chapter on cognitive psychology, be- nius, it comes out as an insult.
cause it assembles criticisms from philosophy Staddon’s attack, like his portrayal of Skin-
and artificial intelligence. The presentation ner’s view of theory, is often distorted and is
of theoretical behaviorism, the next-to-last much too harsh. He misrepresents Skinner’s
chapter, is difficult to interpret, because Stad- views on responsibility, for example, promot-
don introduces the idea of internal states but ing himself as wiser or more humane. Skin-
never clarifies their ontological status. First ner’s objections were to mentalism and au-
he portrays hypothetical constructs as ‘‘place tonomous man, both of which Staddon also
holders’’ in quantitative models. That sounds would reject, but in a discussion parallel to
all right; state variables are useful in models. the one in my own book, Understanding Be-
He goes on, however, to call them ‘‘internal haviorism (Baum, 1994), Staddon redefines
states’’ and to suggest that models tell how responsibility in practical terms, with the dif-
the brain behaves and that the physiologists’ ference that he then uses his redefinition to
job is to ‘‘figure out how the brain can behave argue illogically that Skinner was wrong. Like
76 WILLIAM M. BAUM

Skinner, he argues, too, that freedom is a alone rejection. Staddon is simply trying to
practical affair, mentioning Skinner’s stress tar Skinner with the postmodernist brush so
on feeling free, but calling it subjective, and that he can justify including a critique of rel-
then claiming that Skinner was in some (un- ativism that would otherwise seem out of
explained) way wrong about that, too. place. He calls Barbara Herrnstein Smith
On top of criticism that, if misguided, is at ‘‘Skinner-influenced.’’ She knew Skinner, but
least grounded, Staddon goes on to ridicu- that hardly constitutes grounds for blaming
lous extremes. He includes a long discussion her excesses on Skinner, and Staddon offers
of the necessity and virtues of punishment in no evidence.
social policy that rests entirely on the unex- As I read through his diatribe against Skin-
amined assumption that punishment works as ner and Skinnerians, I wondered why Stad-
a deterrent. I was reminded of an article don would write this. What purpose could it
(Staddon, 1995) that he published in the At- serve? Most likely, it is a political move; he
lantic Monthly magazine some years ago, in wants to distance himself from Skinner and
which he presented similar arguments in fa- to curry favor with the anti-Skinner psychol-
vor of the death penalty, completely ignoring ogists and philosophers. This fails, however,
that all research so far indicates it is ineffec- because those folks rarely read or understand
tive as a deterrent. (Perhaps he should have Skinner’s writings, and Staddon winds up sim-
followed his own advice about premature ex- ply strengthening their prejudices. The hard-
trapolation.) The capper, however, was his er but more honorable way would have been
treatment of Barbara Herrnstein Smith, a to take from Skinner what is valuable while
postmodern social constructionist well de- acknowledging the source.
serving of criticism. I enjoyed Staddon’s crit- Skinner, in fact, wrote much that was valu-
icism of postmodernism and Smith, even able for a science of behavior. In what ap-
though it seemed like going off on a tangent. pears as an anomalous page in which he ac-
He points out the illogic of the central prem- knowledges Skinner’s contributions, Staddon
ise that there is no such thing as truth. (Is it calls him a ‘‘brilliant experimenter’’ and
true or false?) He responds to the accusation mentions that ‘‘Skinner provided a concep-
that scientific theories are often judged by tual framework for understanding learning
other than objective standards. (So what?) I that (I believe) has yet to be fully explored,’’
was distressed, however, that he tries to blame although he can’t help adding, ‘‘even though
Smith’s excesses on Skinner. Worse, he does his strictures against theory prevented him
this by innuendo and conjunction. First, we from exploiting it himself and impeded the
read that Skinner’s ideas about truth ‘‘have efforts of others to do so’’ (Staddon, 2001a,
taken root in some strange lands, most re- p. 122). This is damning with faint praise.
cently in the politically correct thickets of lit- Skinner invented the ‘‘Skinner box,’’ but his
erary theory’’ (Staddon, 2001a, p. 67). Fancy experiments were much less significant than
prose, this, but are we supposed to believe what he found to say about them. The science
that Skinner was responsible for political cor- we practice today would have been impossible
rectness? Then he tells us that Skinner is ‘‘cit- without Skinner’s invention of the concepts
ed with approval by more than one fan of the of operant and stimulus control and his em-
‘postmodern aesthetic,’ ’’ as if being cited phasis on rate as a dependent variable. Stad-
were a sin. Staddon goes on to say that the don himself owes a huge intellectual debt to
views of ‘‘Skinner and Skinnerians’’ (note the Skinner, if he would but consider. If Staddon
insertion of fellow travelers) resemble those has seen far (and that is an open question),
of ‘‘postmodernists/deconstructionists and it is because he stood on the shoulders of gi-
relativists such as Foucault, Derrida, and La- ants, one of whom was Skinner and another
tour’’ in that all ‘‘are skeptical of rationalism, a student of Skinner’s.
objectivity, and the idea of an independent, In the included preface to the first edition,
external reality’’ (Staddon, 2001a, p. 80). As which was a shorter and much less anti-Skin-
someone who has searched long and hard for ner book, Staddon writes, ‘‘I never used to
evidence of Skinner’s views about external re- think of myself as a behaviorist, but now I see
ality, I would challenge anyone to find sup- that I have been ignoring the evidence’’
port in his writings even for skepticism, let (Staddon, 2001a, p. xv). On the occasion of
 A REVIEW OF THE NEW BEHAVIORISM 77

writing, he discovered himself to be a behav- to find: Staddon embraces that very idea. He
iorist. Would that he had embraced this re- tells us first that theoretical models incorpo-
alization. Instead, he resists it and writes as rate internal states and second that ‘‘These
an outsider, referring to behaviorists as models are the behavior . . . what the organ-
‘‘they’’ and disparaging their institutions. ism is ‘doing,’ described in the most color-
Their journals, he reports, publish research less, direct way possible’’ (Staddon, 2001a, p.
‘‘in the Skinnerian tradition’’ (Staddon, 144). To me, this statement seems indistin-
2001a, p. 122). From the context, we are left guishable from the cognitivists’ competence-
in no doubt that this includes the Journal of performance distinction. How would devia-
the Experimental Analysis of Behavior, in which tions of observed behavior from the model’s
Staddon himself has often published. The im- predictions be interpreted? After that, Stad-
plication appears to be that behavior analysts don brings out the example that appears in
(presumably excepting Staddon) have made previous publications: cascaded leaky integra-
no conceptual advances beyond Skinner’s tors as a model of habituation. He shows how
framework. This is self-aggrandizing and a model composed of two cascaded integra-
false. About research on choice since Herrn- tors predicts rate dependence. Curiously,
stein’s (1961) discovery of the matching law, however, he fits the model to no data, and
Staddon comments, ‘‘. . . a whole industry the only data shown are grouped—the pro-
arose to study the topic of choice’’ (p. 39), portion of a group of nematodes responding
thus to denigrate and dismiss decades of to a tap on the container—whereas the mod-
thoughtful experimentation as plodding and el predicts strength of response in an individ-
directionless (the implication of ‘‘industry’’), ual organism. This is suggestive, but hardly
presumably because he thinks it made no the- compelling.
oretical advance. Perhaps if Staddon made The final chapter on consciousness is the
more effort to notice the research of other best, although only loosely connected to the
behavior analysts, he would be less inclined rest of the book. Staddon explains the ‘‘Tur-
to present it so falsely. ing-test view of consciousness: If a subject can
The positive contribution of the book lies make the appropriate verbal report, we will
in a section of chapter 6 (‘‘Mind and Mech- grant him consciousness’’ (Staddon, 2001a,
anism’’) on theoretical behaviorism and pp. 161–162). He goes on to say, ‘‘Theoretical
chapters 7 (‘‘Internal States’’) and 8 (‘‘Con- behaviorism accepts the Turing-test view of
sciousness and Theoretical Behaviorism’’). consciousness: If you act conscious (as as-
The presentation of theoretical behaviorism sessed by an admittedly fallible human in-
in chapters 6 and 7 takes up only 19 pages quisitor), you are conscious’’ (p. 162). Unfor-
(less than 10% of the book). In it, Staddon tunately, he never explains why and never
makes the transition from state variables to acknowledges the significance of ‘‘admittedly
internal states without justification and leaves fallible.’’ Never mind, I liked it anyway. Stad-
the reader in confusion as to exactly what he don leaves off his axe grinding and discusses
intends about their status. In chapter 6, he consciousness in a way consistent with other
attacks cognitivism on two grounds: philoso- contemporary behavioristic treatments (e.g.,
pher John Searle’s (1992) criticism of the Baum, 1994; Rachlin, 1994). He presents and
brain-computer analogy, and research on ar- uses a three-part distinction among the do-
tificial intelligence that demonstrates the mains of experience (the subjective, about
emergent intelligence of many unintelligent which science has nothing to say), physiology
units operating in concert, implying the lack (the functioning of the brain, about which
of need for internal or mental representa- much is still to be learned), and behavioral
tions. I welcomed the criticisms of the com- data (intersubjectively verifiable reports).
puter analogy and representations, but no- Armed with this, he debunks several antibe-
ticed that Staddon carefully avoided havioral and indeed antiscientific arguments
criticizing the antibehavioral aspect of cog- based on consciousness as a thing and as a
nitivism: The denial of behavior as a subject cause.
matter in preference to an idea that behavior I am sorry I could find so little positive to
is only the evidence of inner processes that say about this book. I found its negative tone
are the real subject matter. The reason is easy offensive, its presentation of the place of the-
78 WILLIAM M. BAUM

ory unclear and ambiguous, and much of the Ferster, C. B., & Skinner, B. F. (1957). Schedules of rein-
forcement. New York: Appleton-Century-Crofts.
philosophical discussion imprecise and half- Herrnstein, R. J. (1961). Relative and absolute strength
baked. This was true despite my agreement of response as a function of frequency of reinforce-
with most of the essential points. In particu- ment. Journal of the Experimental Analysis of Behavior, 4,
lar, I agree that behavior analysis has devel- 267–272.
Herrnstein, R. J. (1970). On the law of effect. Journal of
oped to the point where it is ready for theory, the Experimental Analysis of Behavior, 13, 243–266.
as Skinner suggested it would. Maturity in a Herrnstein, R. J. (1974). Formal properties of the match-
science may be gauged by the degree to ing law. Journal of the Experimental Analysis of Behavior,
21, 159–164.
which its data are organized by the applica- Mach, E. (1960). The science of mechanics: A critical and
tion of mathematics. Models like Staddon’s historical account of its development. Lasalle, IL: Open
allow the development of mathematical treat- Court. (Translation of the ninth German edition,
ments of data. That is their value. Who reifies 1933. Original work published 1883.)
Rachlin, H. (1994). Behavior and mind: The roots of modern
the variables in a mathematical model by call- psychology. New York: Oxford University Press.
ing them internal states makes a mistake. I Ryle, G. (1949). The concept of mind. Chicago: University
notice that Staddon has another book on the- of Chicago Press.
Searle, J. (1992). The rediscovery of the mind. Cambridge,
ory (2001b), Adaptive Dynamics: The Theoretical MA: MIT/Bradford.
Analysis of Behavior. I look forward to seeing Skinner, B. F. (1948). ‘‘Superstition’’ in the pigeon. Jour-
the theories without the polemics. nal of Experimental Psychology, 38, 168–172. Reprinted
in Catania, A. C. (1968). Contemporary research in oper-
ant behavior (pp. 62–64). Glenview, IL: Scott, Fores-
man and Company.
REFERENCES Skinner, B. F. (1950). Are theories of learning necessary?
Psychological Review, 57, 193–216. Reprinted in Cata-
Baum, W. M. (1974). On two types of deviation from the nia, A. C. (1968) Contemporary research in operant behav-
matching law: Bias and undermatching. Journal of the ior (pp. 4–21). Glenview, IL: Scott, Foresman and
Experimental Analysis of Behavior, 22, 231–242. Company.
Skinner, B. F. (1986). Some thoughts about the future.
Baum, W. M. (1979). Matching, undermatching, and
Journal of the Experimental Analysis of Behavior, 45, 229–
overmatching in studies of choice. Journal of the Exper- 235.
imental Analysis of Behavior, 32, 269–281. Staddon, J. (1995). On responsibility and punishment.
Baum, W. M. (1994). Understanding Behaviorism: Science, Atlantic Monthly, 275(2), 88–94.
Behavior, and Culture. New York: HarperCollins. Staddon, J. E. R. (2001a). The new behaviorism. Philadel-
Davison, M., & Baum, W. M. (2000). Choice in a variable phia: Taylor & Francis.
environment: Every reinforcer counts. Journal of the Staddon, J. E. R. (2001b). Adaptive dynamics. Cambridge,
Experimental Analysis of Behavior, 74, 1–24. MA: MIT/Bradford.