2001 baum-molar versus molecular as

7/31/2019 2001 Baum-molar Versus Molecular As

1/4

338 COMMENTARY

arise from several sources (Sidman, 1966),not necessarily all at the same time but selec-tively, given the context. I believe avoidancebehavior can be reinforced by the termina-tion of external, internal, or response-pro-

duced stimuli that have been closely correlat-ed with shock, by escape from behavior thathas been closely paired with shock, by the re-duction of shock density, and now, by the pro-duction of a safe period. For me, Dinsmoorhas ruled out none of these possibilities, but

with the response-produced safe period, hehas added a powerful and perhaps more

widely applicable explanatory principle to theothers that are available.

REFERENCES

Dinsmoor, J. A. (2001). Stimuli inevitably generated bybehavior that avoids electric shock are inherently re-

inforcing. Journal of the Experimental Analysis of Behavior,75, 311333.

Dinsmoor, J. A., & Sears, G. W. (1973). Control of avoid-

ance by a response-produced stimulus. Learning andMotivation, 4, 284293.

Herrnstein, R. J. (1970). On the law of effect. Journal ofthe Experimental Analysis of Behavior, 13, 243266.

Herrnstein, R. J., & Hineline, P. N. (1966). Negative re-inforcement as shock-frequency reduction. Journal ofthe Experimental Analysis of Behavior, 9, 421430.

Schoenfeld, W. N. (1950). An experimental approach toanxiety, escape and avoidance behavior. In P. H. Hoch& J. Zubin (Eds.), Anxiety (pp. 7099). New York:Grune & Stratton.

Sidman, M. (1953a). Avoidance conditioning with briefshock and no exteroceptive warning signal. Science,118, 157158.

Sidman, M. (1953b). Two temporal parameters of themaintenance of avoidance behavior by the white rat.

Journal of Comparative and Physiological Psychology, 46,253261.

Sidman, M. (1962). Reduction of shock frequency as re-inforcement for avoidance behavior. Journal of the Ex-

perimental Analysis of Behavior, 5, 247257.Sidman, M. (1966). Avoidance behavior. In W. Honig

(Ed.), Operant behavior: Areas of research and application(pp. 448498). New York: Appleton-Century-Crofts.

Received November 27, 2000Final acceptance December 8, 2000

MOLAR VERSUS MOLECULAR ASA PARADIGM CLASH

WILLIAM M. BAU MUNIVERSITY OF NEW HAMPSHIRE

The molar view of behavior arose in response to the demonstrated inadequacy of explanations basedon contiguity. Although Dinsmoors (2001) modifications to two-factor theory render it irrefutable,a more basic criticism arises when we see that the molar and molecular views differ paradigmatically.The molar view has proven more productive.

Key words: molar view, molecular view, contiguity, atomism, two-factor theory, paradigm

Behavior analysis inherited from 19th-cen-tury psychology an atomistic view of behaviorand environment. Although we no longer

talk about the association of ideas, the termsstimulus and responseare still with us. Hand inhand with this atomism went the principle of

Correspondence should be addressed to the author,611 Mason #504, San Francisco, California 94108 (E-mail:

[email protected]).

association by contiguity, which moved byanalogy from classical conditioning to instru-mental and operant conditioning (Baum,

1995). As a principle of association or rein-forcement, contiguity served to get the sci-ence going, but eventually showed itself to beinsufficient. Dinsmoor (2001) defends 19th-century atomism against the onslaught of anew conceptual framework that arose in the1960s and 1970s. For the present discussion,
http://www.ncbi.nlm.nih.gov/entrez/query.fcgimailto:[email protected]:[email protected]://www.ncbi.nlm.nih.gov/entrez/query.fcgi


2/4

339COMMENTARY

I will refer to 19th-century atomism based oncontiguity as the molecular view and the new-er framework, based on extended patternsand relations, as the molar view.

Rescorla (1967, 1968), in both experimentsand arguments, clarified the explanatory in-sufficiency of contiguity. Contiguity of theconditional stimulus with the unconditionalstimulus cannot and does not predict classicalconditioning, because the unconditionalstimulus must also be less frequent in the ab-sence of the conditional stimulus. If the un-conditional stimulus is just as likely to occurin the absence of the conditional stimulus asin its presence, no conditioning occurs, de-spite the contiguity of the stimuli. AlthoughRescorla made the point for classical condi-tioning, it is readily extended to operant con-ditioning (Bloomfield, 1972). If the reinforc-

er is just as available in the absence ofresponding as in the presence of responding,we expect no operant responding. In molec-ular termsthat is, in the terms of occur-rence and nonoccurrencethe time periodsbefore and after a response must differ in thefrequency of the reinforcer.

Occurrence versus nonoccurrence is justthe crudest and most obvious way in which areinforcer or punisher may vary. Creatures

will behave so as to enhance the efficacy of areinforcer or diminish the intensity of a pun-isher. They will behave so as to bring a rein-forcer nearer in time or to delay a punisher.

They also will behave so as to increase thelikelihoodhence, the rateof a reinforcerand to decrease the likelihood (rate) of apunisher.

The molar view is thus an extension of thenecessary comparison that invalidates thecontiguity-based law of effect. Suppose, how-ever, one granted the invalidation of conti-guity while still insisting on a molecular view,as Dinsmoor (2001) apparently does. Whatadvantage lies in the molar view? Dinsmoorsattempt to defend two-factor theory illustrates

well the advantage, because his argumentsrender his theory irrefutable and redundant.

Let us agree that behavior produces pro-prioceptive and kinesthetic feedback. It israrely observed, but Dinsmoor (2001) has asolution to that: Assume the stimuli and ob-serve the response. His explanation of theavoidance found by Herrnstein and Hineline(1966), with no exteroceptive stimulus and

random delivery of shocks both before andafter responses, is that the response-producedstimuli are less often paired with shock thanpreresponse (i.e., postshock) stimuli. Thetranslation, taking into account that the re-sponse is the measure of the stimuli, is thatshock occurs at a higher rate before respons-es than after. That, however, is the shock-ratereduction explanation. Dinsmoor cannot ex-plain the avoidance without reference toshock-rate reduction, even though he prefersto talk about delay, which he himself recog-nizes is simply the reciprocal of rate. Thestatement, In the absence of responding,the time intervals from one shock to the nextare shorter on average, is exactly equivalentto the statement, In the absence of respond-ing, the shock rate is higher.

Dinsmoor (2001) insists that the response-

produced stimuli, which seem redundant, infact are essential, because reinforcementmust be immediate. Behavior is maintainedless well when unsignaled delays are intro-duced between reinforcers and the responsesthat produce them. I discussed this in 1973,because it seems to imply a role for contiguity

within the context of the molar view. I sug-gested that the effect of unsignaled delays isto degrade the correlation between rate ofreinforcement and response rate, and Ishowed some sample records from an exper-iment testing this idea. Dinsmoor ignoresthat part of the article and the feedback func-

tion, the essential ingredient in appreciatingthe organismenvironment feedback system.Ironically, after insisting on the necessity ofimmediate consequences, Dinsmoor tries toexplain away the maintenance of respondingthat only eventually, after a minimum of 2min, shortens the avoidance session, by sug-gesting that the delayed escape reinforces theresponding (Mellitz, Hineline, Whitehouse,& Laurence, 1983). This looks like trying tohave it both ways.

I said before, and I say again, that two-fac-tor theory cannot explain avoidance withoutresorting to hypothetical entities (Baum,

1973, 1989). The hypothetical entity in Dins-moors (2001) attempted explanation is thereinforcement. His appeals to aversivenessand safety are no more defensible than wasMowrers (1960) appeal to fear. We knowthat under certain circumstances creatures

will behave so as to avoid electric shock. To


3/4

340 COMMENTARY

say that the reason for this is the aversive-ness of the shock is to add nothing to theaccount. It is exactly like saying that objectsare heavy because they possess letharge, arehot because they possess caloric, or burn be-cause they possess phlogiston. Dinsmoorneeds this imaginary essence only because heneeds something to transfer from shock tosignal. A parallel point applies to safety andto conditioned reinforcement, where theimaginary essence might be called reinforc-ingness.

The experiments that Dinsmoor (2001)cites as evidence for his theory are concep-tually flawed, because they depend on failureof discrimination. They are fragile results, be-cause they depend on a host of factors, anyone of which might facilitate or prevent theformation of the discrimination of the lack of

contingency or shock. In the experiment byWeisman and Litner (1969), for example,Dinsmoor omits to mention that the rats werepretrained to avoid shock. No doubt thereare many ways to confuse a rat.

The conflict here, however, is not betweentheories, as Dinsmoor (2001) seems to sug-gest. The conflict is paradigmatic. Dinsmoorsdefense of two-factor theory should be readas a defense of an atomistic view of behavioranalysis. He shows that two-factor theory canexplain anything, much as followers of Ptol-emy, in defense of the geocentric view of thesolar system, showed that epicycles could ex-

plain all the perturbations in the paths of theplanets. Copernicuss heliocentric view actu-ally predicted the planets motions no betterthan the geocentric view, but it prevailed inthe end, because it was the more productive

view. Similarly, the molar view of behavioranalysis has proven to be the more productive

view.What Dinsmoor (2001) takes to be a weak-

ness of the molar view is actually a strength.There is no question that extended relationsmay be overridden by local relations. Hu-mans and other animals sometimes behave soas to obtain immediate small reinforcers even

when by doing so they cancel the possibilityof later larger or more frequent reinforcers.They also sometimes behave so as to post-pone immediate small punishers even whenby doing so they produce later larger or morefrequent punishers. The molar view allowsthese observations to be cast in terms that

may be studied: Under what conditions doshort-term small reinforcers exert more con-trol than later large reinforcers? More gen-erally, under what conditions do local rein-forcement and punishment prevail overextended reinforcement and punishment?The question ties the overriding to a largebody of research on self-control (Logue,1995; Rachlin, 2000). Pigeons behavior, likethat of humans, sometimes is controlled bylocal relations and sometimes by extended re-lations (e.g., Rachlin & Green, 1972). Re-search is beginning to help us to understand

why (Rachlin, 2000).Likewise, the molar view casts the effects of

delay into questions for research. The pas-sage of time affects behavior in at least two

ways. Stimuli and responses affect present be-havior less and less as they recede into the

past, and events that will occur sometime inthe future, even when their delay is signaled,affect present behavior less and less the moreremote they are. These effects are probablyrelated. Research on timing, delayed discrim-ination, and temporal discounting all cometogether to focus on this problem, which maybe called the problem of time horizonthatis, the problem of discovering the factors thatdetermine loss and gain of effectiveness withthe passage of time.

Dinsmoor (2001) seems to think that theproblem of time horizon is somehow fatal forthe molar view, because he thinks that when

one calculates a response rate, the time pe-riod in the denominator is arbitrary. His notrecognizing that the question of time horizonis an empirical question arises from his mo-lecular view. To him, response rate is a con-

venient summary, a derived measure that in-dicates response strength or probability. Incontrast, the molar view takes response rateas a real entity, an aspect of a pattern of be-havior that is extended in time. Extended pat-terns are composed of more local patterns,and every extended pattern is part of somestill more extended pattern (Baum, 1995,1997). A single key peck may be part of a

pattern of pecking at a keysay, variable-in-terval performancethat in turn may be partof a pattern of choice between two keys, thatin turn may be part of a pattern of behavioralallocation between key pecking and otherfood-related behavior, and so on.

The idea that behavioral patterns are nec-


4/4

341COMMENTARY

essarily extended in time, on which the molarview is based, raises an insoluble problem forthe molecular view (Baum, 1997). Dinsmoor(2001) takes the measurement of the re-sponse for granted. He overlooks that, whena lever is depressed, one knows if it was areal lever press only if one eventually (i.e.,after a while) comes to see it as part of anextended pattern. The rat might havestepped on the lever while exploring or fallenon the lever while sniffing at the ceiling. Thenotion of operant level was an attempt to deal

with this point, but it fails, because more isinvolved. For example, in Schneiders (1969)classic study of fixed-interval performance,he had to devise a way to assign some re-sponses to an initial low-rate period and otherresponses to the later high-rate period. Hedid this by analyzing the pattern of respond-

ing as a whole.In closing, I would like to set out somequestions for Dinsmoor. Not that I think he

will lack answers, but rather that I think hisanswers will be revealing. First, if the re-sponse-produced proprioceptive stimuli arereinforcing, then they reinforce responsescontinuously. Why does free-operant avoid-ance responding occur only at moderaterates, instead of at high rates characteristic ofcontinuous reinforcement?

Next, how does the molecular view explainthe differences between ratio and interval re-sponse rates? This is really a two-part ques-

tion. I expect that Dinsmoor would say thatrate on interval schedules is lower than onratio schedules because of differential rein-forcement of long interresponse times on in-terval schedules. Two problems arise. First,differential reinforcement of long interre-sponse times should eventually reduce re-sponse rate to the point at which the averageinterresponse time would be long enoughthat probability of reinforcement would beabout 1.0. Response rate should be lowenough that the schedule would approximatecontinuous reinforcement. Response ratesmaintained by interval schedules are always

much higher than this. Why is response rateon interval schedules so high? Second, why isresponse rate on ratio schedules so high?Here, differential reinforcement of interre-sponse times cannot apply, because all inter-response times are reinforced with equalprobability. One suggestion was that ratio

schedules reinforce bursts of responses, butthat raises the question of how one definesburst without referring to response rate; for

what is a burst but a period of unusually highrate? How will Dinsmoor construct an expla-nation that differs from the molar explana-tion: that ratio schedules differentially rein-force high response rates with high rates ofreinforcement?

REFERENCESBaum, W. M. (1973). The correlation-based law of effect.

Journal of the Experimental Analysis of Behavior, 2 0, 137153.

Baum, W. M. (1989). Quantitative prediction and molardescription of the environment. The Behavior Analyst,12, 167176.

Baum, W. M. (1995). Introduction to molar behavioranalysis. Mexican Journal of Behavior Analysis, 21, 725.

Baum, W. M. (1997). The trouble with time. In L. J.Hayes & P. M. Ghezzi (Eds.), Investigations in behavioralepistemology(pp. 4759). Reno, NV: Context Press.

Bloomfield, T. M. (1972). Reinforcement schedules:Contingency or contiguity? In R. M. Gilbert & J. R.Millenson (Eds.), Reinforcement: Behavioral analyses(pp.165208). New York: Academic Press.

Dinsmoor, J. A. (2001). Stimuli inevitably generated bybehavior that avoids electric shock are inherently re-inforcing. Journal of the Experimental Analysis of Behavior,75, 311333.

Herrnstein, R. J., & Hineline, P. N. (1966). Negative re-inforcement as shock-frequency reduction. Journal ofthe Experimental Analysis of Behavior, 9, 421430.

Logue, A. W. (1995). Self-control: Waiting until tomorrowfor what you want today. Englewood Cliffs, NJ: PrenticeHall.

Mellitz, M., Hineline, P. N., Whitehouse, W. G., & Laur-ence, M. T. (1983). Duration-reduction of avoidancesessions as negative reinforcement. Journal of the Ex-

perimental Analysis of Behavior, 40, 5767.Mowrer, O. H. (1960). Learning theory and behavior. New

York: Wiley.Rachlin, H. (2000). The science of self-control. Cambridge,

MA: Harvard University Press.Rachlin, H., & Green, L. (1972). Commitment, choice,

and self-control. Journal of the Experimental Analysis of Behavior, 17, 1522.

Rescorla, R. A. (1967). Pavlovian conditioning and itsproper control procedures. Psychological Review, 74,7180.

Rescorla, R. A. (1968). Probability of shock in the pres-ence and absence of CS in fear conditioning. Journalof Comparative and Physiological Pyschology, 66, 15.

Schneider, B. A. (1969). A two-state analysis of fixed-interval responding in the pigeon. Journal of the Ex-

perimental Analysis of Behavior, 12, 677687.Weisman, R. G., & Litner, J. S. (1969). Positive condi-

tioned reinforcement of Sidman avoidance behaviorin rats. Journal of Comparative and Physiological Psychol-ogy, 68, 597603.

Received November 15, 2000Final acceptance December 1 2, 2000
http://www.ncbi.nlm.nih.gov/entrez/query.fcgihttp://www.ncbi.nlm.nih.gov/entrez/query.fcgi

2001 baum-molar versus molecular as

Documents