Douglas A. Williams

Configural and elemental strategies in predictive learning

When 2 cues occur together and reliably predict an outcome, Ss often judge the effect of the compound as reducible to the individual effects of the elements. This elemental processing in predictive learning is perhaps the single most important aspect of most theories of human inference. Surprisingly, selectional processing was not observed in either blocking or conditioned inhibition problems. Only when the learner had past experience with another problem encouraging an elemental strategy were the expected selectional processes observed. These proactive effects of prior learning were abolished if the earlier problem required a nonadditive solution. The results suggest that configural cues were guiding predictive inferences in the absence of elemental processes

FORMS OF INHIBITION IN ANIMAL AND HUMAN LEARNING

Forms of inhibition were identified in human predictive learning that are qualitatively similar to those identified by P.C. Holland (1984) in rats. When P (positive) signaled the outcome and PN (N = negative) signaled the absence of the outcome, participants learned the discrimination, but the negative cue did not suppress responding to a transfer cue. Post-learning reversal training, in which N was followed by the outcome, did not abolish the original discrimination. These 2 results imply a configural form of inhibition. Negative transfer, which indicated a 2nd, elemental form of inhibition, was observed when neither PN nor N were reinforced during the discrimination stage. Under these conditions, negative transfer and the original discrimination were both abolished by individually pairing N with the outcome. Empirical parallels and differences with the animal conditioning literature are discussed.

Timed excitatory conditioning under zero and negative contingencies.

Rats (rattus norvegicus) anticipated the arrival of a food pellet unconditioned stimulus (US) even when the conditioned stimulus (CS) signaled no overall change or a substantial decrease in the overall rate of US occurrence. Pellet USs were scheduled probabilistically in the intertrial interval at either an equivalent rate (Experiment 1) or a four times higher rate (Experiments 2 and 3) than in the CS, which included one fixed-time target US. Conditioning has been said to involve learning whether (contingency) the CS signals a change in the US, and if so, when (contiguity) the US is scheduled to arrive. Our results suggest that when trumps whether, challenging the received view that a positive CS-US contingency is necessary for successful conditioning.

Associative and normative accounts of negative transfer.

Subjects were given a prediction task in which they had to learn that one cue, P (positive), was followed by the outcome, and a compound of P and another cue, N (negative), was not followed by the outcome. Next, N was tested in compound with a transfer cue, T, which had signalled the outcome but had never been compounded with N. Experiment 1 confirmed an important assumption of the Rescorla–Wagner model (Wagner & Rescorla, 1972) that negation of T should depend on the specific P cue compound with N being positively contingent. Experiments 2 and 3 confirmed the models prediction that no decrement in negative transfer should be observed following postlearning devaluation of P. Unfortunately, the model did not anticipate that a large proportion of devaluation trials relative to learning trials would attenuate negative transfer (Experiment 4), nor did it predict that negative transfer would occur when P was neutral during the learning stage and was only later made positive (Experiment 3). The results can be accommodated by the Rescorla–Wagner model if one assumes that absent cues have their associative strengths altered.

Elemental and configural processing of novel cues in deterministic and probabilistic tasks

Abstract Three experiments examined whether transfer of past learning depends on how well the original discrimination is learned. To vary terminal learning levels, cues were either perfect (deterministic) or imperfect (probabilistic) predictors of a trial’s outcome. Participants in Experiment 1 acquired a configural, an elemental or a control discrimination. Tests for generalization showed that past learning influenced the processing of new compounds formed from elements of the original discrimination, especially so when the original discrimination was deterministic. Similar results were found in Experiment 2 when the test stimuli were elements derived from novel compounds presented after the original discrimination was acquired. Experiment 3 used filler trials to equate learning in the probabilistic and deterministic tasks, and demonstrated that final levels of learning, rather than task per se, was the critical variable mediating transfer. Implications for rule-based and associative theories are discussed.

Acquired equivalence learning with antecedent and consequent unconditioned stimuli.

The conditions under which cues with similar training histories are treated by rats as being equivalent were examined. A significant event, food, either preceded, followed, or preceded and followed each of the Stimuli A and B; Stimulus C was not food-associated. Next, B was counterconditioned with footshock until it suppressed an appetitively motivated behavior. When A and C were tested for generalized suppression, A (treated like B) evoked more suppression than C (treated differently than B). However, this equivalence effect was strong only when A and B predicted food and were treated alike.

Conditioning across the duration of a backward conditioned stimulus.

Five conditioned suppression experiments examined the extent to which an appetitively motivated lever-press response can be punished by different components of a backward conditioned stimulus (CS). Using a 0-s unconditioned stimulus (US)-CS interval, Experiments 1 and 2 showed that the initial 3 s of a normally 30-s backward CS served as a more effective punisher than the CS as a whole. Experiment 3 found no such effect if the US-CS interval were 3 s rather than 0 s. Experiments 4A and 4B found that if the US-CS interval were 0 s, the initial part of the backward CS acquired excitatory properties although the CS as a whole passed a summation test for conditioned inhibition. By contrast, the 3-s US-CS interval supported inhibitory conditioning across the whole duration of the backward CS. Taken together, these findings support a modified version of Wagners sometimes opponent process model, which suggests that different components of a backward CS become either excitatory or inhibitory depending on the components temporal proximity to the US.

Mechanisms of second-order conditioning with a backward conditioned stimulus.

Five conditioned suppression experiments with rats examined the conditions under which backward pairings endow a first-order conditioned stimulus (CS1) with the ability to serve as a secondary reinforcer. Experiments 2-5B found evidence for excitatory second-order conditioning (SOC) if, during first-order pairings, the US-CS1 interval was 0 s rather than 3 s. Levels of SOC were comparable after forward and backward pairings (Experiments 1-3), and were unaffected by extinction of CS1 after SOC (Experiment 3). These results suggest that forward and backward CS1s support SOC for the same reason, and they call into question the need to invoke any special mechanism such as memory integration.

Input coding in animal and human associative learning

Two appetitive conditioning experiments with rats investigated whether the degree of generalization between a compound and its component parts is fixed or variable. Both experiments used a two-stage transfer design. In Stage 1, the elemental groups learned that a compound and its component parts signaled the same outcome (i.e. C+, D+, CD+), whereas the configural groups learned that a compound and its component parts signaled different outcomes (i.e. C+, D+, CD-, where + is pellets and - is no pellets). In Stage 2, the rats were tested for reductions in generalization. Experiment 1 found no evidence that past configural learning reduced generalization when a new set of alike-treated A and B elements were presented in compound for the first time. Experiment 2 found no evidence that past configural learning reduced generalization when the stimuli of Stage 1 were presented in a new C-, D-, CD+ relation. In contrast to findings with humans, these results suggest that past experience plays a minor role in how stimuli are encoded in animal conditioning.

Timing during inhibitory conditioning.

Analyses of second-by-second conditioned responding into a food receptacle by hungry rats (Rattus norvegicus) found that inhibition varies across the duration of a conditioned stimulus (CS) in a manner consistent with initial training. Variations in the arrival time of the unconditioned stimulus (US) supported temporally specific suppression of responding (Experiment 1). Summation and retardation tests (Experiments 2 and 3, respectively) revealed that points of greatest inhibition coincided with US omission at the time normally specified by the excitor. Our data provide a clear demonstration of fine-grained changes in the time course of inhibitory conditioning for the first time.