Effects of similarity and experience on discrimination learning: a nonassociative connectionist model of perceptual learning.

This article describes a novel connectionist model of perceptual learning (PL) that provides a mechanism for nonassociative differentiation (J. J. Gibson & E. J. Gibson, 1955). The model begins with the assumption that 2 processes--1 that decreases associability and 1 that increases discriminability--operate during preexposure (S. Channell & G. Hall, 1981). In contrast to other models (e.g., I. P. L. McLaren, H. Kaye, & N. J. Mackintosh, 1989), in the current model the mechanisms for these processes are compatible with a configural model of associative learning. A set of simulations demonstrates that the present model can account for critical PL phenomena such as exposure learning and effects of similarity on discrimination. It is also shown that the model can explain the paradoxical result that preexposure to stimuli can either facilitate or impair subsequent discrimination learning. Predictions made by the model are discussed in relation to extant theories of PL.     

Pattern structure and relational discrimination learning.

McGonigle and Jones take exception to Dodwell's explanation of anomalous transfer (AT) as the outcome of relational discriminations among a set of stimulus patterns which vary along a single dimension of orientational salience, from 'horizontal' to 'vertical'. In particular they do not think that the continuum is generated by units with Hubel-and-Wiesel type retinal receptive fields. Instead, they invoke Garner's notions about stimulus structure to explain Dodwell's results, as well as their own finding of two situations where AT fails to occur. It seems that McGonigle and Jones missed the point of the relational discrimination explanation of AT. In fact, it is shown that AT is not predicted by this model for the conditions in which they failed to obtain it. The relational model makes definite predictions about conditions under which AT will occur; as this is not true of their invocation of Garner's ideas, the former is to be preferred. Whether or not outputs are coded by Hubel-and Wiesel type units is not particularly relevant to the main point, that AT is a result of relational learning. In fact, the model is an instance of how structure, in Garner's sense, can be generated.     

ALCOVE: an exemplar-based connectionist model of category learning.

ALCOVE (attention learning covering map) is a connectionist model of category learning that incorporates an exemplar-based representation (Medin & Schaffer, 1978; Nosofsky, 1986) with error-driven learning (Gluck & Bower, 1988; Rumelhart, Hinton, & Williams, 1986). Alcove selectively attends to relevant stimulus dimensions, is sensitive to correlated dimensions, can account for a form of base-rate neglect, does not suffer catastrophic forgetting, and can exhibit 3-stage (U-shaped) learning of high-frequency exceptions to rules, whereas such effects are not easily accounted for by models using other combinations of representation and learning method.     

Prejudiced learning: A connectionist account

Connectionist simulation was employed to investigate processes that may underlie the relationships between prior expectancies or prejudices and the acquisition of attitudes, under conditions where learners can only discover the valence of attitude objects through directly experiencing them. We compared contexts analogous to learners holding either false negative expectancies (‘prejudices’) about a subclass of objects that were actually good or false positive expectancies about objects that were actually bad. We introduced expectancy‐related bias either by altering the probability of approach, or by varying the rate of learning following experience with good or bad objects. Where feedback was contingent on approach, the false positive expectancies were corrected by experience, but negative prejudices resisted change, since the network avoided objects deemed to be bad, and so received less corrective feedback. These findings are discussed in relation to the effects of intergroup contact and expectancy‐confirmation processes in reducing or sustaining prejudice.     

Combining exemplar-based category representations and connectionist learning rules.

Adaptive network and exemplar-similarity models were compared on their ability to predict category learning and transfer data. An exemplar-based network (Kruschke, 1990a, 1990b, 1992) that combines key aspects of both modeling approaches was also tested. The exemplar-based network incorporates an exemplar-based category representation in which exemplars become associated to categories through the same error-driven, interactive learning rules that are assumed in standard adaptive networks. Experiment 1, which partially replicated and extended the probabilistic classification learning paradigm of Gluck and Bower (1988a), demonstrated the importance of an error-driven learning rule. Experiment 2, which extended the classification learning paradigm of Medin and Schaffer (1978) that discriminated between exemplar and prototype models, demonstrated the importance of an exemplar-based category representation. Only the exemplar-based network accounted for all the major qualitative phenomena; it also achieved good quantitative predictions of the learning and transfer data in both experiments.     

Expectancy confirmation in attitude learning: A connectionist account

Connectionist computer simulation was employed to model how learners acquire attitudes towards novel objects under conditions where (a) they are given prior expectancies that the objects as a whole are mostly good or mostly bad; and (b) they can only discover the true valence of the objects by approaching them. Expectancy confirmation was operationalized through modifying connection weights more after experiencing good than bad objects (positive bias), or more after experiencing bad than good objects (negative bias). Negative bias led the network to misclassify more good objects as bad, such negative attitudes resisting change because of the lack of corrective feedback relating to avoided objects. Conversely, positive bias encouraged approach and hence feedback leading to more accurate discrimination of good and bad objects, as well as to higher estimates of the valence of objects not presented during training. These findings suggest that expectancy confirmation may emerge “automatically” from basic learning processes. Copyright © 2008 John Wiley & Sons, Ltd.     

Developments in cognitive neuroscience: II. Implications for theories of transference

An integration of psychoanalytic theory with contemporary developments in cognitive neuroscience offers a useful perspective on long-standing controversies about the nature of transference, and a better understanding of the precise mechanisms by which transferential processes occur. Contemporary psychoanalytic views of transference are reviewed, and the many processes that constitute transference are described. Two issues that have emerged in different guises for several decades-the role of the analyst in eliciting transference, and the nature of "real" and "transferential" components of the therapeutic relationship-are reconsidered in the light of concepts such as connectionist networks. Although a useful analytic stance is one that allows the patient's enduring dynamics to dominate the analytic field, it is suggested, anonymity is neither a cognitive possibility nor the driving force behind most transference reactions, and the distinction between "real" and "transferential" perceptions is one of therapeutic interest, not of mechanism. Certain features of the analytic situation make some dynamics more likely than others to enter the treatment relationship, notably those related to authority, intimacy and attachment, and sexuality. Transference reactions are best understood as constructed from a combination of the patient's enduring dispositions to react in particular ways under particular conditions; features of the analytic situation and of the analyst; and interactions between patient and analyst. These reactions do not unfold ineluctably from the patient's mind in the consulting room, nor are they cognitive constructions of the patient-analyst dyad or co-constructions of relatively equal partners exerting their influence on the analytic field.     

A configural analysis for feature-negative discrimination learning.

In 4 experiments, rats were initially trained with an A+ AXo discrimination in which Stimulus A by itself signaled the delivery of food, A+, whereas the simultaneous presentation of A and X was followed by nothing, AXo. In each experiment X was then paired with food prior to a test phase in which A and X were again presented for a discrimination. The discrimination was of the form A+ AXo in Experiments 1 and 2, whereas it was of the form X+ AXo for Experiments 3 and 4. In all 4 experiments the test discrimination was acquired more rapidly than a control discrimination. The results are interpreted in terms of the original A+ AXo discrimination resulting in the growth of an association between a representation of the entire AX compound and the effects of nonreinforcement.     

Psychophysical theories of duration discrimination

There are few quantitative theories of duration discrimination and few established empirical phenomena to guide theorizing. This paper discusses three such theories and several empirical findings. The theories assume that the discrimination is based only upon information extracted from the temporal extent of the stimulus pattern, and experimental evidence is presented that clearly supports this assumption for many stimulus patterns. Recent findings which indicate that duration information is analyzed in certain ways that are fundamentally different from other stimulus dimensions are reviewed, the duration discrimination psychometric function is examined, and the time-order error is discussed. The three theories are compared in terms of their ability to incorporate the empirical data.     

The modulation of inhibition of return by objectinternal structure: Implications for theories of object-based attentional selection

Recently, Vecera, Behrmann, and McGoldrick (2000), using a divided-attention task, reported that targets are detected more accurately when they occur on the same structural part of an object, suggesting that attention can be directed toward object-internal features. We present converging evidence using the object-based inhibition of return (IOR) paradigm as an implicit measure of selection. The results show that IOR is attenuated when cues and targets appear on the same part of an object relative to when they are separated by a part boundary. These findings suggest that object-based mechanisms of selection can operate over shape representations that make explicit information about objectinternal structure.     

Discrimination of structure: II. Feature binding

In 3 experiments, pigeons acquired a discrimination between patterns comprising the same features. Thus vertical green bars beside horizontal red bars might have signaled food, and horizontal green bars beside vertical red bars might have signaled no food. The solution of this discrimination can be explained by assuming each pattern is represented either by a template or by structural features that are sensitive to combinations of color and line orientation. The 1st explanation predicts subjects should react to a training pattern rotated 90 degrees in the same way as the pattern on which it is based. The 2nd explanation predicts these patterns should be treated as if they signal opposing outcomes. The experiments confirmed the 2nd of these predictions.     

Category vs. item learning: Implications for categorization models

A paradigm that required that subjects learn two responses to each of 10 schematic faces was used to study the relative rate of discrimination and generalization learning. One response uniquely identified each face, whereas the second response classified each face as a member of one of two categories. Rapid category learning and slow item learning suggested that category responses were learned on the basis of abstracted information, but item responses depended on the more difficult task of discriminating among patterns. The results are related to categorization models and to task variables that should influence the relative rates of discrimination and generalization learning.     

Implicit theories of relationships: Implications for relationship satisfaction and longevity

The role of implicit theories in romantic relationships was investigated in two studies. People holding a soulmate theory, who believed that finding the right person is most important for a satisfying relationship, were compared to people holding a work–it–out theory, who believed that effort is most important for building a successful relationship. In Study 1, college students (N = 527) completed a set of questionnaires, including measures of relationship theories and functioning within romantic relationships. Approximately 8 months later, a subset of these students (N = 176) completed a second set of questionnaires for Study 1. The implicit theories were highly stable over time (r = .74). For soulmate theorists, feelings that one’s specific partner is ideal predicted relationship satisfaction and relationship longevity to a greater extent than for work–it–out theorists. Whereas Study 1 investigated people’s theories of relationships as ends of a bipolar continuum, a separate study explored people’s theories of relationships as two unipolar dimensions. Findings from Study 2 (N = 266) supported a representation of the theories as two negatively correlated factors, and supported findings from Study 1 highlighting the role of the interaction between the relationship theories and partner fit in predicting relationship satisfaction.     

Implications of goal theories for the theories of reasoned action and planned behaviour

This paper argues that the predictive validity of the theories of reasoned action and planned behaviour could be enhanced by considering key ideas from goal theories. In particular, goal theories highlight the importance of: (a) construing action as a process of behavioural selection designed to achieve actors’ goals, (b) assessing the extent to which people have planned how to perform action sequences implied by their goals (c) investigating goal conflict in order to understand intention-behaviour discrepancies (d) examining contextual variations in goal salience to account for the dynamics of choice, (e) using intention stability to index the prioritization of goals, and (f) analyzing the content of the goals underlying attitudes and intentions. Suggestions are made about self-report measures and computations that would permit greater use of these ideas in future research.