Esther Mondragón

Analysis of the perceptual learning effect in flavour aversion learning: Evidence for stimulus differentiation

Rats received exposure to two compound flavours, AX and BX, where A and B were sucrose and saline and X was acid. For group intermixed (I), exposure consisted of alternating trials with AX and BX; group blocked (B) received a block of AX trials and a separate block of BX trials. Experiment 1 showed that generalization to BX after conditioning with AX was less profound in group I than in group B. Separate examination of the elements of the compound showed that the source of this difference lay in the strength acquired by the X element. X acquired less strength in group I than in group B (Experiments 1 and 2), whereas for the A element (Experiments 3 and 4) the reverse pattern was obtained. These results support the proposal that the perceptual learning effect (restricted generalization from AX to BX in group I) depends on a process that enhances the effectiveness of unique stimulus elements (A and B) and reduces that of common elements (such as X).

Making the illusory correlation effect appear and then disappear: The effects of increased learning

The acquisition of a negative evaluation of a fictitious minority social group in spite of the absence of any objective correlation between group membership and negative behaviours was described by Hamilton and Gifford (1976) as an instance of an illusory correlation. We studied the acquisition and attenuation through time of this correlation learning effect. In two experiments we asked for participants’ judgements of two fictitious groups using an online version of a group membership belief paradigm. We tested how judgements of the two groups changed as a function of the amount of training they received. Results suggest that the perception of the illusory correlation effect is initially absent, emerges with intermediate amounts of absolute experience, but diminishes and is eliminated with increased experience. This illusory correlation effect can be considered to reflect incomplete learning rather than a bias due to information loss in judgements or distinctiveness.

Computational Neuroscience for Advancing Artificial Intelligence: Models, Methods and Applications

tools to analyze the structure of psychological models. But they are just abstract tools after all. In any empirical science, the ultimate proof rests on experimental evidence. Nonetheless, perhaps paradoxically, here it is precisely where the full strength of symmetries shows: Not from the models of theories built on symmetry principles but from the intimate connection (through symmetry arguments) between such models and observed phenomena. If we look back to the problems faced by psychological models of associative learning as listed in section 2, we find that they relate to deficiencies that symmetry could be used to resolve. The first shortcoming, that no model accounts for all associative learning phenomena, refers to a lack of explanatory power in such models; the second one, that contradictory rules explain the same phenomena, claims for a normative approach; the third one, that models are partial, relates to the need for unifying principles where different theories that cover disjoint phenomena find common grounds and are made compatible; and the fourth one, that some phenomena remain unaccounted for, identifies a classification problem. It seems, therefore, that symmetries may be useful in solving such problems. First we must find the psychological symmetries. This is the purpose of our research. 6. IN SEARCH OF PSYCHOLOGICAL SYMMETRIES Although there is not a universally accepted ‘law of learning’, all psychological models coincide in assuming that learning takes place when a (relatively permanent) change in behavior happens as a consequence of some experience. Now, we need to know whether such law establishes sufficient symmetry conditions for the occurrence of the observed phenomena –or, in other words, we have to investigate whether the observed phenomena describe necessary conditions for the law to hold (invariantly) true. Unfortunately, a glimpse at the literature suggests it does not: 1. That the sensory and motivational features of the stimuli as well as their novelty and relevance affect learning are well documented facts (Kamin and Schaub, 1963; Pavlov, 1927; Jenkins and Moore, 1973; Randich and LoLordo, 1979; Lubow, 1989; Garcia and Koelling, 1966); 2. Procedurally, the idea that learning is contextspecific is also gaining ground (Bouton, 1993; Bouton and Swartzentruber, 1986; Hall and Mondragón, 1998); also, different results emerge depending on the order in which stimuli are presented during training and on the number (single or compound) and representation (elemental or configural) of the cues themselves (see, e.g., Pearce and Bouton, 2001 for a survey). This first setback may not challenge our search for psychological symmetries though. It could we argued that, after all, we should expect that the parameters in (a) affected the pace of learning (accelerating or decelerating the learning process, i.e., strengthening or weakening the links between nodes/stimuli as time goes), defining, in the extreme, explicit symmetry breaks. Unfortunately, the study of complex phenomena in (b) does not only tell us that the learning rate changes in different experimental conditions. What these results tell us is that the rules of learning themselves fluctuate depending on such factors and, consequently, that they do not reflect any genuine object of invariance. Not surprisingly, a mathematical analysis of the above-mentioned issues reveals that each of them violates one of the conditions for group formation: Associativity. This is rather worrying since associativity is the key condition for symmetry. It tells us that the concatenation of two different operations gives the same result, and that gives us

Serial order of conditional stimuli as a discriminative cue for Pavlovian conditioning.

The serial order in which events occur can be a signal for different outcomes and therefore might be a determinant of how an animal should respond. In this report, we propose a novel design for studying serial order learning in Pavlovian conditioning. In both Experiments 1a and 1b, hungry rats were trained with successively presented pairs of auditory and visual stimuli (e.g., A --> B) using four different stimuli (A-D). Four orders were paired with food (A --> B, B --> C, C --> D, D --> A) while the reversals were extinguished (B --> A, C --> B, D --> C, A --> D). An analysis of responding from the second element of each pair showed that the rats discriminated trial types that preceded food from those that did not. A replication of the effect using a completely counterbalanced design is described in Experiment 1b. These results suggest that rats can use the serial or temporal order of two sequentially presented non-overlapping elements as the basis for discrimination. Two associative accounts are suggested as possible mechanisms for solving the discrimination.

The Effect of Stimulus Distribution Form on the Acquisition and Rate of Conditioned Responding: Implications for Theory

In four experiments rats were conditioned to an auditory conditioned stimulus (conditioned stimulus; CS) that was paired with food, and learning about the CS was compared across two conditions in which the mean duration of the CS was equated. In one, the CS was of a single, fixed duration on every trial, and in the other the CS duration was drawn from an exponential distribution, and hence changed from trial to trial. Higher rates of conditioned responding to the fixed than to the variable stimulus were observed, in both between- (Experiment 1) and within-subject designs (Experiments 2 and 3). Moreover, this difference was maintained when stimuli trained with fixed or variable durations were tested under identical conditions (i.e., with equal numbers of fixed and variable duration trials)-suggesting that the difference could not be attributed to performance effects (Experiment 3). In order to estimate the speed of acquisition of conditioned responding, the scaled cumulative distribution of a Weibull function was fitted to the trial-by-trial response rates for each rat. In the within-subject experiments specific differences in the pattern of acquisition to fixed and variable CS were shown; a somewhat different pattern was found when intertrial interval (ITI) was manipulated (Experiment 4). The implications of these findings for theories of conditioning and timing are discussed.

Perceptual learning in an appetitive Pavlovian procedure: analysis of the effectiveness of the common element.

Non-reinforced preexposure to two stimuli often enhances discrimination between them. Analyses of this perceptual learning phenomenon have mainly focused on the role played by the distinctive stimulus features; this study examined the contribution of the non-distinctive common elements. A standard appetitive Pavlovian procedure was used. Rats received two different schedules of exposure--alternated or blocked--to two compound auditory stimuli, AX and BX. In Experiment 1 a generalization test to BX that followed conditioning to AX showed that animals responded less, and hence discriminated better, following alternated exposure, thus extending the generality of this perceptual learning effect to standard appetitive Pavlovian procedures. The degree to which the common element X was mediating this effect was explored in the next three experiments. Experiment 2 assessed the effectiveness of X following conditioning to AX. Experiment 3 explored Xs effectiveness throughout extensive conditioning to X. Experiment 4 tested the ability of X to overshadow a novel stimulus Y. The results were consistent with the suggestion that alternated preexposure can reduce the relative effectiveness of the common element.

Rats do learn XYX rules

Please cite this article in press as: Mondrago In a study carried out with prelinguistic infants, Marcus et al. (1999) proposed that the XYX sequence-learning paradigm constitutes evidence of abstract rule learning related to language that is exclusive to humans. Hauser et al. (2002b) found that cottontop tamarins, Sanguinus oedipus, were also able to learn the sequence and extract a rule, extending the ability from humans to primates. Murphy et al.’s (2008) experiment 2 showed that rats were also able to discriminate the pattern XYX and to transfer it to novel stimuli. Corballis’s (in press) article argues against these later results and questions the involvement of this type of learning in human language. Corballis claims that our rats, and presumably the argument would extend to human infants and cottontop tamarins, confronted with the same kind of task, may have used a subset of stimuli to solve the rule discrimination. For instance, rats learning that XYX was the reinforced sequence may have matched the identity of the first and last stimulus (X), ignoring the interposed element (Y), and that this would be sufficient to discriminate XYX from YYX or YXX. Corballis’s account is not as parsimonious as he suggests because it requires that the rat not only identify each stimulus but also its order or position in the sequence. Moreover, rats must learn which

SSCC TD: A Serial and Simultaneous Configural-Cue Compound Stimuli Representation for Temporal Difference Learning

This paper presents a novel representational framework for the Temporal Difference (TD) model of learning, which allows the computation of configural stimuli – cumulative compounds of stimuli that generate perceptual emergents known as configural cues. This Simultaneous and Serial Configural-cue Compound Stimuli Temporal Difference model (SSCC TD) can model both simultaneous and serial stimulus compounds, as well as compounds including the experimental context. This modification significantly broadens the range of phenomena which the TD paradigm can explain, and allows it to predict phenomena which traditional TD solutions cannot, particularly effects that depend on compound stimuli functioning as a whole, such as pattern learning and serial structural discriminations, and context-related effects.

Overshadowing by fixed- and variable-duration stimuli

Two experiments investigated the effect of the temporal distribution form of a stimulus on its ability to produce an overshadowing effect. The overshadowing stimuli were either of the same duration on every trial, or of a variable duration drawn from an exponential distribution with the same mean duration as that of the fixed stimulus. Both experiments provided evidence that a variable-duration stimulus was less effective than a fixed-duration cue at overshadowing conditioning to a target conditioned stimulus (CS); moreover, this effect was independent of whether the overshadowed CS was fixed or variable. The findings presented here are consistent with the idea that the strength of the association between CS and unconditioned stimulus (US) is, in part, determined by the temporal distribution form of the CS. These results are discussed in terms of time-accumulation and trial-based theories of conditioning and timing.

A Complete Serial Compound Temporal Difference Simulator for Compound stimuli, Configural cues and Context representation

Temporal Difference (TD) (Sutton and Barto, 1987) is a real-time error correction model in which learning is computed according to the difference between successive predictions and a discount factor that decays exponentially, reflecting the fact that predictors closer to a reinforcer (the unconditioned stimulus, US) are based on more recent information and thus more accurate. In addition, an eligibility trace modulates the extent to which the stimulus predictive value is susceptible of changing on any given time-step. The way stimuli are represented affects significantly how learning is implemented in TD. The Complete Serial Compound representation (CSC) (Moore et al., 1998) has become standard in studies of dopamine function (Schultz 2010) and is central in investigating reward-based models of schizophrenia (Smith et al., 2006). This CSC representation is at the core of the TD Simulator that we briefly describe in this item. The simulator has been built upon the graphical interface of the R&W Simulator significantly modified to introduce temporal parameters and constraints. CSC TD assumes that a stimulus can be broken down into a series of individual elements, which are each active for a single unit of time as shown Fig. 1, left. Each of these new elements has a separate eligibility trace and associative strength, receives distinct reinforcement from the US, and while active contributes to the prediction term. In essence CSC TD treats the components of a stimulus as unique stimuli in their own right identifiable by the overarching stimulus and their position in the sequence. These component stimuli are linked only in that their activation is contingent on the activation of the supra-stimulus and their position in the sequence of time-steps. Briefly, the first component of the CSC stimulus becomes active when the supra-stimulus becomes active, and the active state of each component in the sequence is then the product of the active state of the preceding stimulus and that of the suprastimulus. This representation produces correct predictions at the intra-trial level, because elements of the stimulus occurring distantly from the US receive correspondingly less reinforcement, modulated by their eligibility trace as shown in Fig. 1, right. CSC TD is expressed formally as follows: