Generalizing a neuropsychological model of visual categorization to auditory categorization of vowels

Twelve male listeners categorized 54 synthetic vowel stimuli that varied in second and third formant frequency on a Bark scale into the American English vowel categories [see text]. A neuropsychologically plausible model of categorization in the visual domain, the Striatal Pattern Classifier (SPC; Ashby & Waldron, 1999), is generalized to the auditory domain and applied separately to the data from each observer. Performance of the SPC is compared with that of the successful Normal A Posteriori Probability model (NAPP; Nearey, 1990; Nearey & Hogan, 1986) of auditory categorization. A version of the SPC that assumed piece-wise linear response region partitions provided a better account of the data than the SPC that assumed linear partitions, and was indistinguishable from a version that assumed quadratic response region partitions. A version of the NAPP model that assumed nonlinear response regions was superior to the NAPP model with linear partitions. The best fitting SPC provided a good account of each observer's data but was outperformed by the best fitting NAPP model. Implications for bridging the gap between the domains of visual and auditory categorization are discussed.     

Augmentation in contingency learning under time pressure

Recent research suggests that cue competition effects in human contingency learning, such as blocking, are due to higher‐order cognitive processes. Moreover, some experimental reports suggest that the effect opposite to blocking, augmentation, could occur in experimental preparations that preclude the intervention of reasoning mechanisms. In the present research, we tested this hypothesis by investigating cue interaction effects in an experimental task in which participants had to enter their responses under time pressure. The results show that under these conditions, augmentation, instead of blocking, is observed.     

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.     

Generalization and similarity in exemplar models of categorization: Insights from machine learning

Exemplar theories of categorization depend on similarity for explaining subjects’ ability to generalize to new stimuli. A major criticism of exemplar theories concerns their lack of abstraction mechanisms and thus, seemingly, of generalization ability. Here, we use insights from machine learning to demonstrate that exemplar models can actually generalize very well. Kernel methods in machine learning are akin to exemplar models and are very successful in real-world applications. Their generalization performance depends crucially on the chosen similarity measure. Although similarity plays an important role in describing generalization behavior, it is not the only factor that controls generalization performance. In machine learning, kernel methods are often combined with regularization techniques in order to ensure good generalization. These same techniques are easily incorporated in exemplar models. We show that the generalized context model (Nosofsky, 1986) and ALCOVE (Kruschke, 1992) are closely related to a statistical model called kernel logistic regression. We argue that generalization is central to the enterprise of understanding categorization behavior, and we suggest some ways in which insights from machine learning can offer guidance.     

Temporal constraints on visual learning: a computational model

Given a constant stream of perceptual stimuli, how can the underlying invariances associated with a given input be learned? One approach consists of using generic truths about the spatiotemporal structure of the physical world as constraints on the types of quantities learned. The learning methodology employed here embodies one such truth: that perceptually salient properties (such as stereo disparity) tend to vary smoothly over time. Unfortunately, the units of an artificial neural network tend to encode superficial image properties, such as individual grey-level pixel values, which vary rapidly over time. However, if the states of units are constrained to vary slowly, then the network is forced to learn a smoothly varying function of the training data. We implemented this temporal-smoothness constraint in a backpropagation network which learned stereo disparity from random-dot stereograms. Temporal smoothness was formalized with the use of regularization theory by modifying the standard cost function minimised during training of a network. Temporal smoothness was found to be similar to other techniques for improving generalisation, such as early stopping and weight decay. However, in contrast to these, the theoretical underpinnings of temporal smoothing are intimately related to fundamental characteristics of the physical world. Results are discussed in terms of regularization theory and the physically realistic assumptions upon which temporal smoothing is based.     

Fear makes you stronger: Responding to feared animal targets in visual search

To investigate whether fear affects the strength with which responses are made, 12 animal-fearful individuals (five snake fearful and seven spider fearful) were instructed to decide as quickly as possible whether an animal target from a deviant category was present in a 3 × 4 item (animal) search array. The animal categories were snakes, spiders, and cats. Response force was measured, in newtons. The results showed that the strength of the response was greater when the feared animal served as the target than when it served as the distractors. This finding was corroborated by evoked heart rate changes to the stimuli. Our findings strengthen the argument that focused attention on a single, feared animal can lead to increases in manual force.     

RAP: a new framework for visual categorization

Cognitive science might almost be defined as several disciplines communicating their different perspectives on the mind, the common object of study. However, domain-specific concepts and techniques can prevent, rather than foster, a communication of viewpoints. In this article, we develop a new framework for visual categorization in which the interaction between Represented (R) information and Available (A) information determine the Potent (P) information (symbolically, R ? A approximately P). We argue and illustrate that this framework helps to establish a common language to articulate issues common to low-, mid-, and high-level vision. More importantly, we present new techniques with which to visualize the so-far elusive constructs of representation and potent information.     

The knowledge acquired during artificial grammar learning: Testing the predictions of two connectionist models

An artificial grammar learning experiment is reported which investigated whether three types of information are learned during this kind of task: information about the positions of single letters, about fragments of training strings, and about entire training strings. Results indicate that participants primarily learned information about string fragments and, to a lesser extent, information about positions of letters. Two connectionist models, an autoassociator and a simple recurrent network (SRN), were tested on their ability to account for these results. In the autoassociator simulations, similarity of test items to entire training items had a large effect, which was at variance with the experimental results. The results of the SRN simulations almost perfectly matched the experimental ones. Received: 22 September 1998 / Accepted: 22 February 1999     

Decision making under time pressure: An independent test of sequential sampling models

Choice probability and choice response time data from a risk-taking decision-making task were compared with predictions made by a sequential sampling model. The behavioral data, consistent with the model, showed that participants were less likely to take an action as risk levels increased, and that time pressure did not have a uniform effect on choice probability. Under time pressure, participants were more conservative at the lower risk levels but were more prone to take risks at the higher levels of risk. This crossover interaction reflected a reduction of the threshold within a single decision strategy rather than a switching of decision strategies. Response time data, as predicted by the model, showed that participants took more time to make decisions at the moderate risk levels and that time pressure reduced response time across all risk levels, but particularly at the those risk levels that took longer time with no pressure. Finally, response time data were used to rule out the hypothesis that time pressure effects could be explained by a fast-guess strategy.     

Thirty-something categorization results explained: selective attention, eyetracking, and models of category learning

An eyetracking study testing D. L. Medin and M. M. Schaffer's (1978) 5-4 category structure was conducted. Over 30 studies have shown that the exemplar-based generalized context model (GCM) usually provides a better quantitative account of 5-4 learning data as compared with the prototype model. However, J. D. Smith and J. P. Minda (2000) argued that the GCM is a psychologically implausible account of 5-4 learning because it implies suboptimal attention weights. To test this claim, the authors recorded undergraduates' eye movements while the students learned the 5-4 category structure. Eye fixations matched the attention weights estimated by the GCM but not those of the prototype model. This result confirms that the GCM is a realistic model of the processes involved in learning the 5-4 structure and that learners do not always optimize attention, as commonly supposed. The conditions under which learners are likely to optimize attention during category learning are discussed.     

A mathematical model of learning under schedules of interresponse time reinforcement

A model is proposed for free responding under schedules of interresponse time reinforcement. Each response generates a pattern of stimuli that changes in an orderly way over time. At any time the changing pattern may be sampled, and the state of conditioning of the sampled pattern determines whether or not a response actually occurs. The prediction of this model for the asymptotic distribution of interresponse times is derived. The predictions are tested against data from ratio, interval, and DRL schedules of reinforcement. The fit of the model is sufficiently good to make it worth-while continuing investigation of the model.     

When parameters collide: a warning about categorization models

Similarity-choice (S—C) models of categorization contain two principal mathematical transformations: an exponential-decay similarity function and a choice rule. However, there is a tension between the psychological processes that models emulate and the mathematics they use to do so. To illustrate this, I show that in these models an unappreciated interaction occurs between the mathematical transformations so that the stages of the model essentially cancel each other out. The result is that the model’s output reflects its input linearly. This cancellation phenomenon has potentially serious implications regarding the interpretation and use of S—C models. The phenomenon also raises questions about the simplification and psychological grounding of categorization models. Modelers broadly might benefit from an internal analysis of their models, such as that described here.     

Observational learning of two visual discriminations by pigeons: a within-subjects design.

Pigeon's observational learning of successive visual discrimination was studied using within-subject comparisons of data from three experimental conditions. Two pairs of discriminative stimuli were used; each bird was exposed to two of the three experimental conditions, with different pairs of stimuli used in a given bird's two conditions. In one condition, observers were exposed to visual discriminative stimuli only. In a second condition, subjects were exposed to a randomly alternating sequence of two stimuli where the one that would subsequently be used as S+ was paired with the operation of the grain magazine. In a third experimental condition, subjects were exposed to the performance of a conspecific in the operant discrimination procedure. After exposures to conspecific performances, there was facilitation of discriminative learning, relative to that which followed exposures to stimulus and reinforcement sequences or exposures to stimulus sequences alone. Exposure to stimulus and food-delivery sequences enhanced performance relative to exposure to stimulus sequences alone. The differential effects of these three types of exposure were not attributable to order effects or to task difficulty; rather, they clearly were due to the type of exposure.     

Testing sequence effects in visual memory: clues for a structural model

In order to probe the internal organization of visual temporary memory, systematic experiments were performed in which the subjects had to memorize a series of 2-5 images then pass recognition tests, either in all possible testing permutations, (in the case of 2-4 images) or in 20 selected permutations (in the case of 5 images). Over 300,000 tests were performed, generating more than 40,000 errors. The error-rates were found to follow simple rules. Both the 3 and the 4 images results are compatible with the presence of only four typical accuracy levels. On the other hand, the reaction time (RT) results revealed surprisingly rich patterns. The RT for recognizing image i at testing stage t > 1 depends upon which image (j) was tested just before. The ranking of the RTs for (j, i) couples evolves from one testing stage to the next. It is proposed that these RTs reflect, in part, the time needed to localize the trace in memory of a given image, starting from the position at which the previous test occurred. If this assumption is correct, the results of this study are providing a picture of the configurations formed by memorized items in the visual temporary store. A hypothetical minimal model, involving several rows of slots on a triangular mesh is proposed to account for the structures in both the error-rates and the RT results.     

Effects of pre- or post-training paradoxical sleep deprivation on two animal models of learning and memory in mice

The aim of the present study was to verify the effects of pre- or post-training paradoxical sleep (PS) deprivation in mice tested in the passive and the plus-maze discriminative avoidance tasks. Three-month-old Swiss male mice were placed in narrow platforms in a water tank for 72 h to prevent the occurrence of PS. Control animals were kept in the same room, but in their home cages. Before or after this period, the animals were submitted to the training session of one of the behavioral tasks. The test sessions were performed 3 and 10 days after the training. The animals that were PS-deprived before the training session showed retention deficits in the test sessions performed 3 days later in both tasks (decreased latency to enter the dark chamber of the passive avoidance apparatus or increased percent time spent in the aversive arm of the plus-maze discriminative avoidance apparatus). Animals that were PS deprived after the training session showed no differences from control animals in the test sessions performed 3 days after the training in any of the tasks, but showed passive and discriminative avoidance retention deficits in the test performed 10 days after the training. The results suggest that both pre- and post-training paradoxical sleep deprivation produce memory deficits in mice. However, these effects have different temporal characteristics.     

Memory and learning for visual signals in time and space

Vision is often characterized as a spatial sense, but what does that characterization imply about the relative ease of processing visual information distributed over time rather than over space? Three experiments addressed this question, using stimuli comprising random luminances. For some stimuli, individual items were presented sequentially, at 8 Hz; for other stimuli, individual items were presented simultaneously, as horizontal spatial arrays. For temporal sequences, subjects judged whether each of the last four luminances matched the corresponding luminance in the first four; for spatial arrays, they judged whether each of the right-hand four luminances matched the corresponding left-hand luminance. Overall, performance was far better with spatial presentations, even when the entire spatial array was presented for just tens of milliseconds. Experiment 2 demonstrated that there was no gain in performance from combining spatial and temporal information within a single stimulus. In a final experiment, particular spatial arrays or temporal sequences were made to recur intermittently, interspersed among, non-recurring stimuli. Performance improved steadily as particular stimulus exemplars recurred, with spatial and temporal stimuli being learned at equivalent rates. Logistic regression identified several shortcut strategies that subjects may have exploited while performing our task.     

Induction and categorization in young children: a similarity-based model

The authors present a similarity-based model of induction and categorization in young children (SINC). The model suggests that (a). linguistic labels contribute to the perceived similarity of compared entities and (b). categorization and induction are a function of similarity computed over perceptual information and linguistic labels. The model also predicts young children's similarity judgment, induction, and categorization performance under different stimuli and task conditions. Predictions of the model were tested and confirmed in 6 experiments, in which 4- to 5-year-olds performed similarity judgment, induction, and categorization tasks using artificial and real labels (Experiments 1-4) and recognition memory tasks (Experiments 5A and 5B). Results corroborate the similarity-based account of young children's induction and categorization, and they support both qualitative and quantitative predictions of the model.     

Infant temperament and information processing in a visual categorization task

This study investigates the relation between temperament and information processing in 7-month-old infants. Infant temperament was assessed on multiple dimensions, using mothers' ratings on the Infant Behavior Questionnaire (IBQ). Information processing was measured during a paired-comparison visual categorization task. Infants (N=50) were familiarized with five different pairs of visual stimuli from one category, followed by two test trials, each contrasting a novel familiar-category item with a novel out-of-category item (animals vs. vehicles). Drop-out rate in this task was within a normal range (21%). No differences in temperament were found between completers and non-completers, but high reactivity to novelty and low motor activity were positively correlated with the strength of infants' familiarization response. Group comparisons between infants scoring high and low on these two dimensions of temperament confirmed this general finding: infants high on distress to novelty and low on motor activity showed stronger familiarization responses. Furthermore, infants scoring low on motor activity showed a stronger categorization response than highly active infants. This study provides evidence that performance in a cognitive categorization task and dimensions of infant temperament are systematically related during the second half of the first year of life. Implications for cognitive research with infants of this age-range will be discussed.     

Change of preferences under time pressure: choices and judgements

Students made choices between student apartments judged to be equally attractive. When asked to choose the best apartment in each pair of alternatives, choices made by a group under time pressure were significantly different from those made by another group under no time pressure. The effect was so strong that preference reversals were obtained which could be explained by subjects over-evaluating negative aspects on the most important attribute under time pressure.