Formation Of Perceptual Categories Research Articles

A neural ensemble correlation code for sound category identification.

Humans and other animals effortlessly identify natural sounds and categorize them into behaviorally relevant categories. Yet, the acoustic features and neural transformations that enable sound recognition and the formation of perceptual categories are largely unknown. Here, using multichannel neural recordings in the auditory midbrain of unanesthetized female rabbits, we first demonstrate that neural ensemble activity in the auditory midbrain displays highly structured correlations that vary with distinct natural sound stimuli. These stimulus-driven correlations can be used to accurately identify individual sounds using single-response trials, even when the sounds do not differ in their spectral content. Combining neural recordings and an auditory model, we then show how correlations between frequency-organized auditory channels can contribute to discrimination of not just individual sounds but sound categories. For both the model and neural data, spectral and temporal correlations achieved similar categorization performance and appear to contribute equally. Moreover, both the neural and model classifiers achieve their best task performance when they accumulate evidence over a time frame of approximately 1–2 seconds, mirroring human perceptual trends. These results together suggest that time-frequency correlations in sounds may be reflected in the correlations between auditory midbrain ensembles and that these correlations may play an important role in the identification and categorization of natural sounds.

Auditory perceptual category formation does not require perceptual warping.

Categorical perception occurs when a perceiver's stimulus classifications affect their ability to make fine perceptual discriminations and is the most intensively studied form of category learning. On the basis of categorical perception studies, it has been proposed that category learning proceeds by the deformation of an initially homogeneous perceptual space ("perceptual warping"), so that stimuli within the same category are perceived as more similar to each other (more difficult to tell apart) than stimuli that are the same physical distance apart but that belong to different categories. Here, we present a significant counterexample in which robust category learning occurs without these differential perceptual space deformations. Two artificial categories were defined along the dimension of pitch for a perceptually unfamiliar, multidimensional class of sounds. A group of participants (selected on the basis of their listening abilities) were trained to sort sounds into these two arbitrary categories. Category formation, verified empirically, was accompanied by a heightened sensitivity along the entire pitch range, as indicated by changes in an EEG index of implicit perceptual distance (mismatch negativity), with no significant resemblance to the local perceptual deformations predicted by categorical perception. This demonstrates that robust categories can be initially formed within a continuous perceptual dimension without perceptual warping. We suggest that perceptual category formation is a flexible, multistage process sequentially combining different types of learning mechanisms rather than a single process with a universal set of behavioral and neural correlates.

Learning novel phonetic categories in perception and production.

Much is known about perceptual category formation in both first and second languages. For example, researchers have demonstrated that in the laboratory listeners are able to form novel perceptual categories after a short period of exposure [e.g., Hayes (2003); Maye & Gerken (2000); Maye & Gerken (2001)]. However, little research has examined how perceptual category formation proceeds in production or how training in one modality affects learning in the other modality. This study systematically manipulates training modality while testing participants in both perception and production. Participants were trained on a novel phonetic category contrast using an implicit learning paradigm [Maye & Gerken (2000); Maye & Gerken (2001)]. Across participants, training either consisted of perception‐only, in which participants did not explicitly produce tokens, or perception+production, in which participants were instructed to explicitly repeat tokens. Participants were tested on their ability to perceive and produce the new contrast before and after training. Perception results demonstrate that the perception‐only training group learns to discriminate the trained contrast. The perception+production training group discriminates less well, suggesting that production training can disrupt perceptual learning. Results of this study will increase understanding of how category formation occurs in both modalities and give insight into the relationship between perception and production.

ERP indices of speech processing in bilinguals

The specific aim of this project was to examine perceptual differentiation of the vowel contrast [I-E] by early versus late bilingual Spanish–English speakers, compared to monolingual English speakers. Neither of these phonetic segments occur as phonemes in Spanish while they constitute phonemic contrast in English. For Spanish listeners, these vowels may be heard as variants of the Spanish phonemes [i] and [e], respectively. The formation of perceptual categories may be affected by exposure to two languages in which the phonemic status of phonetic segments differs. In the current study, electrophysiological measures (mismatched negativity—MMN) and behavioral measures (AX discrimination and identification) were employed to examine perception of a nine-step continuum resynthesized from natural tokens. Preliminary results indicate that early bilinguals show somewhat more categorical perception than monolinguals (i.e., sharper identification boundaries) of the nine-step continuum. MMN results, obtained during a task in which subjects were not attending to the auditory input, show clear MMNs for monolinguals for both cross- and within-category pairs. To date, results for early bilinguals show less clear indication of preattentive discrimination. The role of language experience and the relationship between the behavioral performance and the electrophysiological measures will be discussed.

Generalization in response to mate recognition signals.

Females usually exhibit strong and unequivocal recognition of conspecific mating signals and reject those of other sympatric heterospecifics. However, most species are allopatric with one another, and the degree to which females recognize mating signals of allopatric species is more varied. Such mating signals are often rejected but are sometimes falsely recognized as conspecific. We studied the dynamics of mate recognition in female túngara frogs (Physalaemus pustulosus) in response to a series of calls that were intermediate between the conspecific and each of five allopatric-heterospecific calls: two that elicited recognition from females in previous studies and three that did not. This study shows that females perceive variation in allopatric mating signals in acontinuous manner with no evidence of perceptual category formation. The strength of recognition is predicted by how different the target stimulus is from the conspecific call within a series of calls. But the differences in recognition responses among call series are not predicted by the similarity of the call series to the conspecific call. The latter result suggests that the strength of recognition of allopatric signals might be influenced by processes of stimulus generalization and past evolutionary history.

Animal models of speech sound categorization

Several recent studies have examined phonetic category formation by training nonhuman animals (e.g., birds) to respond differentially to distributions of speech sounds. Animal models provide good tests of theories of perceptual category formation because one can completely control the input distributions and obtain a fine-grained sampling of responses to the stimulus space. In studies using vowel sound distributions, animals’ response structures reveal two salient characteristics. First, there is a tendency to respond most strongly to stimuli that are furthest from the boundaries between training distributions. Second, there is evidence that the structure of responses is a function of the statistics of the training input. This can be seen as a peak in response for stimuli near the centroid of the input distribution. These same response structure characteristics can also be seen in data collected from humans in categorization tasks. It is possible that these two characteristics are indicative of two systems of learning: one functional and one statistical. A similar dichotomy has also been recently proposed in the visual categorization literature. The implications of these results for theoretical and computational models of phonetic acquisition will be discussed. [Work supported by NSF and NIH.]

Auditory-nerve fiber representation of temporal cues to voicing in word-medial stop consonants

The length of the interval between the onset of consonant closure and the onset of voicing in a following vowel is a temporal cue that may distinguish between consonants /d/ and /t/ in word-medial environments; this interval has been called the "consonant duration" [V. W. Zue and M. Laferriere, J. Acoust. Soc. Am. 66, 1039-1050 (1979); S. Davis and W. V. Summers, J. Phon. 17, 339-353 (1989)]. The representation of this cue in the discharge patterns of chinchilla auditory-nerve fibers was measured. The two-syllable utterances /ida/, /ita/, /uda/, and /uta/ were recorded by one male and one female talker. The onset of consonant closure produced discharge rate decreases in nearly all neurons. Either the release of closure or the onset of voicing for the second vowel could elicit an increase in discharge rate. The latencies of these discharge rate changes varied across populations of neurons. A neural measure of consonant duration was extracted from the pattern of latencies. The "encoded duration" was longer for utterances with a medial /t/ than for utterances with a medial /d/. For each utterance the encoded duration increased with increases in characteristic frequency. The variability of the encoded duration measure was small enough to preserve the distinction between utterances with different word-medial consonants. The variability of the encoded duration was large, relative to the acoustic differences between utterances that included the same medial consonant. This pattern of variability could contribute to the formation of perceptual categories by reducing the audibility of within-category acoustic differences.

Learned perceptual categories for complex vocal signals in budgerigars (Melopsittacus undulatus)

Operant conditioning and multidimensional scaling (MDS) procedures were used to study auditory perception of complex vocal signals in budgerigars. Response latencies were anlyzed by MDS to produce a spatial arrangement of these complex sounds reflecting the perceptual organization. Both normal and isolate-reared budgerigars and humans show perceptual categories which correspond closely to the major functional and acoustic categories of species-specific calls. Birds sharing the same learned call show perceptual categories for the calls of individual birds. Humans and other budgerigars unfamiliar with these same learned calls fail to group calls by individual. These results suggest that both general and special auditory mechanisms are involved in the formation of perceptual categories for species-specific vocal signals in budgerigars. Additional tests with human speech sounds suggest that the general auditory processing mechanism in budgerigars are quite sophisticated. [Work supported by NIH.]

Free sorting with stimuli clustered in a multidlmensional attribute space

An experiment was conducted to determine whether Ss leam, without knowledge of results or prior familiarization with the prototypes, to sort histoform stimuli generated by three probablistic rules into categories consistent with these mies. Seven Ss were asked to sort 30 stimuli (10 from each schema population) on each trial. No constraints were placed on the number of categories to be used. A statistic was developed to measure consistency between schema-defined categories and subjeet-defined categories. Four of the seven Ss met the leaming criterion, and demonstrated increasing consistency with the schema-defined categories across trials. In general, more categories than necessary were used to classify the stimuli correctly. The results raise several questions about classification strategies used by Ss in perceptual learning tasks with stimuli generated by probabilistic rules, and indicate the need for further research conceming the variables that influence perceptual category formation.