A feedback model of perceptual learning and categorization

Top-down, feedback, influences are known to have significant effects on visual information processing. Such influences are also likely to affect perceptual learning. This paper employs a computational model of the cortical region inter-actions underlying visual perception to investigate possible influences of top-down information on learning. The results suggest that feedback could bias the way in which perceptual stimuli are categorized and could also facilitate the learning of subordinate level representations suitable for object identification and perceptual expertise.     

基本特征水平的知觉学习与注意的作用

在基本特征水平的知觉学习的研究中是否存在注意与反馈的作用,是一直存在着争论的问题。该研究通过线索不同的有效性来控制注意的分配,探讨了在基本特征水平的知觉学习中注意与反馈的作用,共进行了两个实验,实验一采用内源和外源性线索影响注意分配,探讨基本特征刺激加工过程中注意的作用;实验二对基本特征刺激的知觉学习过程的特性进行研究。学习和探测任务都是让被试从所呈现的刺激中寻找靶子英文字母L。结果发现：在基本特征水平的知觉学习中存在有注意的作用;其学习过程存在反馈的作用。     

The functional form of performance improvements in perceptual learning: learning rates and transfer

The functional form of performance improvements has been extensively studied in speeded cognitive and motor tasks; in such tasks, reductions in response times have been characterized by the ubiquitous power law of learning or by a simpler exponential form. Performance improvements in perceptual capacities are also important in expertise, but their functional form is unknown. This study investigated the functional form of perceptual learning. For individual observers, reductions in thresholds were best described by an exponential function, rather than a power or compound exponential and power (apex) function. Learning was specific to orientation, a result that supports the perceptual locus of the learning, and was decoupled in high and low external noise, a result that reflects separable learning mechanisms in the two conditions. The simple exponential form of learning implies a constant relative rate of learning throughout practice; there was no evidence supporting multilevel hypotheses, such as serial reverse hierarchical and parallel-learning models, that posit multiple processes of learning characterized by different rates.     

Perceptual learning for speech

Adult language users have an enormous amount of experience with speech in their native language. As a result, they have very well-developed processes for categorizing the sounds of speech that they hear. Despite this very high level of experience, recent research has shown that listeners are capable of redeveloping their speech categorization to bring it into alignment with new variation in their speech input. This reorganization of phonetic space is a type of perceptual learning, or recalibration, of speech processes. In this article, we review several recent lines of research on perceptual learning for speech.     

Chunking mechanisms in human learning

Pioneering work in the 1940s and 1950s suggested that the concept of 'chunking' might be important in many processes of perception, learning and cognition in humans and animals. We summarize here the major sources of evidence for chunking mechanisms, and consider how such mechanisms have been implemented in computational models of the learning process. We distinguish two forms of chunking: the first deliberate, under strategic control, and goal-oriented; the second automatic, continuous, and linked to perceptual processes. Recent work with discrimination-network computational models of long- and short-term memory (EPAM/CHREST) has produced a diverse range of applications of perceptual chunking. We focus on recent successes in verbal learning, expert memory, language acquisition and learning multiple representations, to illustrate the implementation and use of chunking mechanisms within contemporary models of human learning.     

Perceptual learning effect on decision and confidence thresholds

Practice can enhance of perceptual sensitivity, a well-known phenomenon called perceptual learning. However, the effect of practice on subjective perception has received little attention. We approach this problem from a visual psychophysics and computational modeling perspective. In a sequence of visual search experiments, subjects significantly increased the ability to detect a “trained target”. Before and after training, subjects performed two psychophysical protocols that parametrically vary the visibility of the “trained target”: an attentional blink and a visual masking task. We found that confidence increased after learning only in the attentional blink task. Despite large differences in some observables and task settings, we identify common mechanisms for decision-making and confidence. Specifically, our behavioral results and computational model suggest that perceptual ability is independent of processing time, indicating that changes in early cortical representations are effective, and learning changes decision criteria to convey choice and confidence.     

Cultural and learning differences in the Judd illusion

In the present study, we examined whether individual differences in the perception of illusory Judd drawings point to variability in the pickup of informational variables. Two sources for these individual differences were addressed: culture and learning. East Asian (n = 24) and Western (n = 24) participants made perceptual judgments of the midpoint of the shaft of various Judd figures in a pretest–practice–posttest design. During practice, half of the participants received feedback about the actual midpoint after each trial, while the other half did not receive feedback. The results showed differences among perceivers of different cultures in judging the midpoints of the shafts of Judd figures, particularly with respect to their propensity to improve perceptual accuracy after repeated practice and feedback. For most participants, changes in illusory bias as a consequence of learning were shown to reflect either a change in what informational variable they exploited or a rescaling or calibration of the perception to the informational variable. However, the individual differences in illusory bias related to culture could not be unequivocally attributed to either of these perceptual-learning processes.     

Comparing lexically guided perceptual learning in younger and older listeners

Numerous studies have shown that younger adults engage in lexically guided perceptual learning in speech perception. Here, we investigated whether older listeners are also able to retune their phonetic category boundaries. More specifically, in this research we tried to answer two questions. First, do older adults show perceptual-learning effects of similar size to those of younger adults? Second, do differences in lexical behavior predict the strength of the perceptual-learning effect? An age group comparison revealed that older listeners do engage in lexically guided perceptual learning, but there were two age-related differences: Younger listeners had a stronger learning effect right after exposure than did older listeners, but the effect was more stable for older than for younger listeners. Moreover, a clear link was shown to exist between individuals’ lexical-decision performance during exposure and the magnitude of their perceptual-learning effects. A subsequent analysis on the results of the older participants revealed that, even within the older participant group, with increasing age the perceptual retuning effect became smaller but also more stable, mirroring the age group comparison results. These results could not be explained by differences in hearing loss. The age effect may be accounted for by decreased flexibility in the adjustment of phoneme categories or by age-related changes in the dynamics of spoken-word recognition, with older adults being more affected by competition from similar-sounding lexical competitors, resulting in less lexical guidance for perceptual retuning. In conclusion, our results clearly show that the speech perception system remains flexible over the life span.     

Tactile learning is task specific but transfers between fingers

Practice-related improvement in visual perception is highly specific for properties of the stimulus used in training. We explored the specificity of such perceptual learning in the human tactile system, using gratings consisting of alternating ridges and grooves. Practice effects on grating discrimination showed limited transfer between grating sets defined by spatial variation in either groove width or ridge width, consistent with partially overlapping neural representations of these two spatial parameters. In contrast, substantial interdigital transfer of practice effects occurred for discrimination of gratings varying in either spatial parameter and also for spatial acuity-dependent discrimination of grating orientation. We conclude that tactile learning, although quite as task specific as in other sensory systems, generalizes with considerable facility across fingers, unlike visual learning, which is highly location specific.     

Direct learning in dynamic touch

A dynamic touch paradigm in which participants judged the lengths of rods and pipes was used to test the D. M. Jacobs and C. F. Michaels (2007) theory of perceptual learning. The theory portrays perception as the exploitation of a locus on an information manifold and learning as continuous movement across that manifold to a new locus, as guided by information available in feedback. The information manifold was defined as a 1-dimensional space of inertial variables. To encourage maximal learning, a 2-step procedure was used in each of 2 experiments. Each step comprised a pretest to identify the starting locus on the information manifold, a practice phase in which feedback specifying the optimal locus was given, and a posttest in which the ending locus on the manifold was identified. In the 2nd step, a different feedback variable specified a different optimum. In both experiments, participants, who sometimes began at different loci, showed the predicted movement toward the optimum in each phase. Whereas previous applications of the theory posit the existence of information-for-learning without identifying a candidate variable, such a candidate is identified.     

Acetylcholine and olfactory perceptual learning

Olfactory perceptual learning is a relatively long-term, learned increase in perceptual acuity, and has been described in both humans and animals. Data from recent electrophysiological studies have indicated that olfactory perceptual learning may be correlated with changes in odorant receptive fields of neurons in the olfactory bulb and piriform cortex. These changes include enhanced representation of the molecular features of familiar odors by mitral cells in the olfactory bulb, and synthetic coding of multiple coincident odorant features into odor objects by cortical neurons. In this paper, data are reviewed that show the critical role of acetylcholine (Ach) in olfactory system function and plasticity, and cholinergic modulation of olfactory perceptual learning at both the behavioral and cortical level.     

Normal and abnormal visual development in kittens: insights into the mechanisms that underlie visual perceptual development in humans

For many years the mechanisms that underlie normal and abnormal development of visual perception in human infants have been explored in anatomical and physiological studies on two species of frontal-eyed mammals, namely, cats and, to a lesser extent, monkeys. The unstated assumption underlying the investigations on cats is that despite substantial differences in the organization of the visual pathways of cats and humans, as well as quantitative differences in their perceptual abilities, principles of development established in the former also apply to humans. This review examines the extent to which this assumption may be valid. Following a review of certain anatomical peculiarities of the cat visual system, several of the differences as well as the parallels between the perceptual abilities of cats and humans are summarized. The latter similarities, as well as the larger number of parallels between the two species that can be drawn during development, attest to the validity of the choice of the cat for study of the mechanisms that underlie human visual development.     

Perceptual learning and reasons-responsiveness

Perceptual experiences are not immediately responsive to reasons. You see a stick submerged in a glass of water as bent no matter how much you know about light refraction. Due to this isolation from reasons, perception is traditionally considered outside the scope of epistemic evaluability as justified or unjustified. Is perception really as independent from reasons as visual illusions make it out to be? I argue no, drawing on psychological evidence from perceptual learning. The flexibility of perceptual learning is a way of responding to new epistemic reasons. The resulting perceptual experiences are epistemically evaluable as justified or unjustified.     

Judgments of learning are significantly higher following feedback on relatively good versus relatively poor trials despite no actual learning differences

Studies have consistently shown that prospective metacognitive judgments of learning are often inaccurate because humans mistakenly interpret current performance levels as valid indices of learning. These metacognitive discrepancies are strongly related to conditions of practice. Here, we examined how the type of feedback (after good versus poor trials) received during practice and awareness (aware versus unaware) of this manipulation affected judgments of learning and actual learning. After each six-trial block, participants received feedback on their three best trials or three worst trials and half of the participants were made explicitly aware of the type of feedback they received while the other half were unaware. Judgments of learning were made at the end of each six-trial block and before the 24-h retention test. Results indicated no motor performance differences between groups in practice or retention; however, receiving feedback on relatively good compared to relatively poor trials resulted in significantly higher judgments of learning in practice and retention, irrespective of awareness. These results suggest that KR on relatively good versus relatively poor trials can have dissociable effects on judgments of learning in the absence of actual learning differences, even when participants are made aware of their feedback manipulation.     

Infant speech perception bootstraps word learning

By their first birthday, infants can understand many spoken words. Research in cognitive development has long focused on the conceptual changes that accompany word learning, but learning new words also entails perceptual sophistication. Several developmental steps are required as infants learn to segment, identify and represent the phonetic forms of spoken words, and map those word forms to different concepts. We review recent research on how infants' perceptual systems unfold in the service of word learning, from initial sensitivity for speech to the learning of language-specific sound patterns. Building on a recent theoretical framework and emerging new methodologies, we show how speech perception is crucial for word learning, and suggest that it bootstraps the development of a separate but parallel phonological system that links sound to meaning.     

Motor learning enhances perceptual judgment: a case for action-perception transfer

Transfer from perception to action is well documented, for instance in the form of observational learning. Transfer from action to perception, on the other hand, has not been researched. Such action-perception transfer (APT) is compatible with several learning theories and has been predicted within the framework of common coding of perceptual and motor events (Prinz, 1992, 1997). Our first experiment aimed at an empirical evaluation of APT and involved motor practice of timed two-cycle arm movements on verbal command without visual feedback. In a transfer test, visual judgments of similar patterns had to be made. In addition, transfer from the visual to the motor task was studied. In Experiment 2 we separated kinesthetic aspects of motor practice from preparatory and efferent contributions to APT. The experiments provide evidence that transfer between perception and action is bi-directional. Transfer from perception to action and, more importantly, from action to perception was found. Furthermore, APT was equally pronounced for participants who had actively practiced movements during training and for passive participants who had received merely kinesthetic feedback about the movement. This kinesthetic-visual transfer is likely to be achieved via visuomotor-kinesthetic matching or via timekeeping mechanisms that are involved in both motor and visual performance. Received: 2 June 1999 / Accepted: 20 June 2000     

The effects of positive versus negative feedback on information-integration category learning

A number of studies have shown that in category learning, providing feedback about errors allows faster learning than providing feedback about correct responses. However, these previous studies used explicit, rule-based tasks in which the category structures could be separated by a simple rule that was easily verbalized. Here, the results of the first experiment known to compare the efficacy of positive versus negative feedback during information-integration category learning are reported. Information-integration tasks require participants to integrate perceptual information from incommensurable dimensions, and evidence suggests that optimal responding recruits procedural learning. The results show that although nearly all of the full-feedback control participants demonstrated information-integration learning, participants receiving either positive-only or negative-only feedback generally used explicit, rule-based strategies. It thus appears that, unlike rule-based learning, consistent information-integration learning requires full feedback. The theoretical implications of these findings for current models of information-integration learning are discussed.     

A mechanism for the direct perception of change: the example of bacterial chemotaxis

The relationship between the behavior of single-celled organisms and cognition in higher animals is explored. Recent research and theory in bacterial chemotaxis are presented, together with a discussion of the implications of chemotaxis for perceptual theory. A number of parallels between chemotaxis and perception in higher organisms are drawn. It is suggested that Koshland's model of the chemical processes controlling chemotaxis is an example of a mechanism for direct perception of change and can help elucidate Runeson's work on 'smart mechanisms' of perception. It is argued, more generally, that the growing body of knowledge about the perceptual activities of lower organisms should be used to broaden the factual base on which theories of perception are constructed: eg explication of perceptual parallels between humans and lower organisms should help clarify the nature of these phenomena in humans and, perhaps, help in the development of theories of greater generality. Also, the debate between direct and indirect theories of perception may be advanced by analysis of the specific mechanisms used by lower organisms. Contrasts to mediated perception are pointed out and arguments for the relative simplicity and explanatory power of theories of direct perception are provided.     

Direct Learning in Auditory Perception: An Information-Space Analysis of Auditory Perceptual Learning of Object Length

The theory of direct learning characterizes perception of a given property as occupying a locus in an information space and characterizes perceptual learning as continuous movement in that information space toward a more optimal locus. Three experiments investigated whether such an information-based account of learning could be applied to perceptual learning in audition. The results of Experiment 1 showed that perception of length by audition could be characterized as occupying a locus in an information space consisting of inertial variables that constrain perception of length by dynamic or effortful touch. Experiments 2 and 3 showed that feedback about length led to predictable movements across the information space from less optimal to more optimal loci. Such results provide additional support for the theory of direct learning and suggest that convergence information may be modality independent.