Academic journal article The Psychological Record

Discrimination Reversal Learning in Capuchin Monkeys (Cebus Apella)

Academic journal article The Psychological Record

Discrimination Reversal Learning in Capuchin Monkeys (Cebus Apella)

Article excerpt

Reports of cognitive abilities in a variety of animal species have surged in recent years (e.g., Wasserman & Zentall, 2006). Many of these reports emerged from studies with nonhuman primates, and attendant explanations for such abilities center on the large brains, complex social groups, and foraging behaviors of these species (e.g., Byrne & Whiten, 1988; Maestripieri, 2003; Rumbaugh & Washburn, 2003). However, much of what we understand about why organisms behave the way that they do comes from conditioning paradigms designed to assess the role of stimulus-response (S-R) associations between stimuli and the extent to which such processes led to generalized forms of learning (Hull, 1943; Spence, 1937).

Thorndike (1911) believed that common associative processes could account for most, if not all, learning in animals and human beings. Although Thorndike recognized that there were important performance differences between species, he believed that these differences were caused by quantitative differences in associative learning processes. Thus, if a monkey could learn a symbolic matching-to-sample task faster than a pigeon, it was because the monkey could form associations faster than a pigeon, not because the monkey was more cognitive than the pigeon. However, Harlow (1949) showed that rhesus monkeys given a large number of novel two-choice discrimination problems in succession showed significant improvement in their acquisition rates. Indeed, after several hundred problems, the monkeys needed only one trial to determine which stimulus was correct and which stimulus was incorrect and could respond correctly on almost all subsequent trials with those same stimuli. Harlow argued that animals that established learning sets were, in effect, learning how to learn the rules that provided for efficient responding during a two-choice discrimination task, where criterional performance on one pair of stimuli led to the introduction of two new stimuli, with each pair containing a randomly determined positive discriminative stimulus (S+) and negative discriminative stimulus (S-). Such increases in efficiency could not result from specific pairings of stimuli and responses (and outcomes) because of the novelty of stimuli across problems, and therefore associative principles could be discounted as accounting solely for the learning that occurred (see Murray & Gaffan, 2006, for a prospective memory account of learning sets).

Rumbaugh (1971) devised a testing paradigm, which he called the mediational paradigm, for differentiating associative and rule-based interpretations for reversal learning in nonhuman animals. The task is a reversal-learning task, in which animals must first learn which of two stimuli is paired with food reward (and which is not) and then learn that the outcomes for selecting the different stimuli had been reversed. The specifics of the task provided different predictions for how animals might respond, depending on whether they were learning through S-R associative processes such as those posited for more general learning phenomena (Hull, 1943; Spence, 1937) or whether they learned about relations between stimuli through the attendant rule learning that accompanies logical task structuring.

In the mediational paradigm, animals first are trained to a criterion of 9 of the last 10 trials correct in a two-choice discrimination (A+B-). Next, the reward contingencies of the two stimuli are reversed for a single A-B+ reversal trial (i.e., the former S+ becomes the S- for this single trial, and the former S- becomes the S+). Then one of three different conditions is presented for an additional 10 trials. In the A-B+ condition, the 10 additional trials are presented with this now reversed contingency. In the B+C- condition, the A stimulus is replaced with a new stimulus (C) that acts as the S- for 10 trials. In the A-C+ condition, the B stimulus is replaced with a new stimulus (C) that acts as the S+ for the additional 10 trials. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.