Baum’s private thoughts

Abstract

I find Baum’s arguments in thisarticle utterly convincing. Thatshould not be surprising, becauseBaum and I think alike. It might beworth explaining why this is so. It hasto do with our common history ofreinforcement and punishment. Weoccupied desks in the same tinycubicle in the basement of Harvard’sMemorial Hall during the 3 years(1962–1965) it took us to get ourPhDs. Together we suffered underthe oppressive system operative there.We stood in awe of Skinner but weremore directly influenced by Herrn-stein and Stevens. To our manydiscussions and near-violent argu-ments, Baum brought his back-ground in biology and I mine inmechanical engineering and gestaltpsychology. These backgrounds gaveus the confidence to focus on evolu-tionary, operational, and molar as-pects of our teachers’ ideas: Skinner’sconcept of the generic nature ofstimulus and response and of theoperant, Stevens’ pragmatic insis-tence on simplicity in theory andexperimental design and, most ofall, Herrnstein’s matching law, themolar implications of which we werejust trying to comprehend.Herrnstein had circulated a work-ing paper applying matching toindividual responses. His originalaccount applied to symmetricalchoices between Response A andResponse B; now he applied thematching equation to a choice be-tween Responses A and not-A (allbehavior except Response A). Ac-cording to Herrnstein (1970), the rateof any particular response was direct-ly proportional to the rate of rein-forcement of that response alone, andwas indirectly proportional to all ofthe reinforcement in the situation(including the reinforcement of thatresponse as well as the reinforcementof all other available behavior). BothBaum and I were bothered by thisaccount. These extra reinforcerscould not be measured directly. Weboth firmly believed in the minimiza-tion of free parameters (we still do),and Herrnstein had just added an-other free parameter to the matchingequation. Also, Herrnstein’s newformulation made a prediction thatwent against our expectations. Sup-pose a pigeon’s pecks were beingreinforced on a variable-intervalschedule (the pigeon consequentlypecking at a steady rate) and thenfree reinforcers, independent of peck-ing, were randomly given to thepigeon (in addition to those it earnedon the variable-interval schedule). Itseemed to us that the extra reinforc-ers, occasionally occurring within ashort time of a peck, might supersti-tiously reinforce pecking and there-fore increase pecking rate. The higherthe rate of these free reinforcers, themore accidental contiguities wouldoccur between pecks and reinforcers,the faster the pigeons should peck.However, Herrnstein’s new formulasaid that pecking rate would varyinversely with the rate of these freereinforcers; the faster they came, theslower the pigeon should peck. Irecall the three of us sitting inHerrnstein’s office as we presentedthis contradiction to him. But he was