Abstract
Humans are able to classify novel items correctly by category1,2; some other animals have also been shown to do this3,4,5,6,7. During category learning, humans group perceptual stimuli by abstracting qualities from similarity relationships of their physical properties1,2,8. Forming categories is fundamental to cognition9 and can be independent of a ‘memory store’ of information about the items or a prototype10. The neurophysiological mechanisms underlying the formation of categories are unknown. Using an animal model of category learning6, in which frequency-modulated tones are distinguished into the categories of ‘rising’ and ‘falling’ modulation, we demonstrate here that the sorting of stimuli into these categories emerges as a sudden change in an animal's learning strategy. Electro-corticographical recording from the auditory cortex11 shows that the transition is accompanied by a change in the dynamics of cortical stimulus representation. We suggest that this dynamic change represents a mechanism underlying the recognition of the abstract quality (or qualities) that defines the categories.
This is a preview of subscription content, access via your institution
Access options
Subscribe to this journal
Receive 51 print issues and online access
$199.00 per year
only $3.90 per issue
Buy this article
- Purchase on Springer Link
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
Similar content being viewed by others
References
Komatsu, L. K. Recent views of conceptual structure. Psychol. Bull. 112, 500–526 (1992).
Medin, D. L. & Coley, J. D. in Perception and Cognition at Century's End: History, Philosophy, Theory (Handbook of Perception and Cognition) (eds Hochberg, J. & Cutting, J. E.) 403–432 (Academic, New York, 1998).
Woodruff, G. & Premack, D. Primitive mathematical concepts in the chimpanzee: proportionality and numerosity. Nature 293, 568–570 (1981).
Herrnstein, R. J. Levels of stimulus control. A functional approach. Cognition 37, 133–166 (1990).
Pearce, J. M. in Current Topics in Animal Learning: Brain, Emotion, and Cognition (eds Dachowski, L. & Flaherty, C.) 141–164 (Erlbaum, Hillsdale, 1991).
Wetzel, W., Wagner, T., Ohl, F. W. & Scheich, H. Categorical discrimination of direction in frequency-modulated tones by Mongolian gerbils. Behav. Brain Res. 91, 29–39 (1998).
Orlov, T., Yakovlev, V., Hochstein, S. & Zohary, E. Macaque monkeys categorize images by their ordinal number. Nature 404, 77–80 (2000).
Feldman, J. Minimization of Boolean complexity in human concept learning. Nature 407, 630–633 (2000).
Smith, E. E. & Jonides, J. in The New Cognitive Neurosciences (ed. Gazzaniga, M. S.) 1013–1022 (MIT Press, Cambridge, Massachusetts, 2000).
Knowlton, B. J. & Squire, L. R. The learning of categories: parallel brain systems for item memory and category knowledge. Science 262, 1747–1749 (1993).
Ohl, F. W., Scheich, H. & Freeman, W. J. Topographic analysis of epidural pure-tone-evoked potentials in gerbil auditory cortex. J. Neurophysiol. 83, 3123–3132 (2000).
Lea, S. E. G. in Animal Cognition (eds Roitblat, H. L., Bever, T. G. & Terrace, H. S.) 263–276 (Erlbaum, Hillsdale, 1984).
Moody, D. B. in Methods in Comparative Psychoacoustics (eds Klump, G. M., Dooling, R. J, Fay, R. R. & Stebbins, W. C.) 293–305 (Birkäuser, Basel, 1995).
Schreiner, C. E., Read, H. L. & Sutter, M. L. Modular organization of frequency integration in primary auditory cortex. Annu. Rev. Neurosci. 23, 501–529 (2000).
Ohl, F. W. & Scheich, H. Orderly cortical representation of vowels based on formant interaction. Proc. Natl Acad. Sci. USA 94, 9440–9444 (1997).
Ohl, F. W., Wetzel, W., Wagner, T., Rech, A. & Scheich, H. Bilateral ablation of auditory cortex in Mongolian gerbil affects discrimination of frequency modulated tones but not of pure tones. Learn. Mem. 6, 347–362 (1999).
Freeman, W. J. Neural mechanisms underlying destabilization of cortex by sensory input. Physica D 75, 151–164 (1994).
Arieli, A., Sterkin, A., Grinvald, A. & Aertsen, A. Dynamics of ongoing activity: explanation of the large variability in evoked cortical responses. Science 273, 1868–1871 (1996).
Heil, P. & Scheich, H. Spatial representation of frequency-modulated signals in the tonotopically organized auditory cortex analogue of the chick. J. Comp. Neurol. 322, 548–565 (1992).
Huotilainen, M. et al. Sustained fields of tones and glides reflect tonotopy of the auditory cortex. NeuroReport 6, 841–844 (1995).
Freeman, W. J. & Schneider, W. Changes in spatial patterns of rabbit olfactory EEG with conditioning to odors. Psychophysiology 19, 44–56 (1982).
Freeman, W. J. & Grajski, K. A. Relation of olfactory EEG to behavior: factor analysis. Behav. Neurosci. 101, 766–777 (1987).
Barrie, J. M., Freeman, W. J. & Lenhart, M. Modulation by discriminative training of spatial patterns of gamma EEG amplitude and phase in neocortex of rabbits. J. Neurophysiol. 76, 520–539 (1996).
Freeman, W. J. Neurodynamics: an Exploration in Mesoscopic Brain Dynamics. (Springer, London, 2000).
Hart, J. Jr, Berndt, R. S. & Caramazza, A. Category-specific naming deficit following cerebral infarction. Nature 316, 439–440 (1985).
McCarthy, R. A. & Warrington, E. K. Evidence for modality-specific meaning systems in the brain. Nature 334, 428–430 (1988).
Creutzfeldt, O., Ojemann, G. & Lettich, E. Neuronal activity in human lateral temporal lobe. I. Responses to speech. Exp. Brain Res. 77, 451–475 (1989).
Martin, A., Wiggs, C. L., Ungerleider, L. G. & Haxby, J. V. Neural correlates of category-specific knowledge. Nature 379, 649–652 (1996).
Kuhl, P. K. & Miller, J. D. Speech perception by the chinchilla: voiced–voiceless distinction in alveolar plosive consonants. Science 190, 69–72 (1975).
Sammon, J. W. Jr A nonlinear mapping for data structure analysis. IEEE Transact. Comp. C-18, 401–409 (1969).
Acknowledgements
We thank K. Buckisch, B. Burke, M. Deliano and D. Labra-Cardero for technical assistance. We also thank J. Altman for critical comments on an earlier version of the manuscript.
Author information
Authors and Affiliations
Corresponding author
Supplementary information
1. Supplementary information on the nature of FM tone discrimination
To address the question of whether the discrimination of rising and falling frequency-modulated (FM) tones might be based on a pitch cue, such as the instantaneous pitch at the beginning or the end of the FM tone, we studied the spontaneous transfer of the conditioned response (CR) to pure tones in gerbils previously trained in the FM tone discrimination paradigm. Four gerbils were trained to discriminate a rising FM tone (2-4 kHz) from a falling FM tone (4-2 kHz). Stimuli and procedure corresponded to training block 1 in the paper. Here the training was stopped after session 15. On the consecutive 3 days (sessions 16-18) presentation of CS+, CS- and US was continued but interspersed were 30 presentations of 2 kHz pure tones and 30 presentations of 4 kHz pure tones (not reinforced) in analogy to the protocol in Ref. 29. The underlying rationale was, that if FM tone discrimination would employ onset or offset pitch cues, elevated levels of CR for the 2 kHz or 4 kHz pure tone, respectively, would be expected. As can be seen in all 4 animals, no spontaneous transfer of the CR occurred in response to the pure tones. This is evidence that FM tone discrimination does not make use of a pitch cue. This interpretation is in accordance with our lesion study (Ref. 16) showing that bilateral lesion of auditory cortex impairs acquisition and retention of FM tone discrimination while having no effect on pure tone discrimination.
This result is implicitly contained in the data presented in the manuscript: The fact that the CR rate in the generalization function (Fig. 1d) for modulation rate = 0 (pure tone) falls to zero indicates that the meaning associated with the trained FM tones is not transferred to pure tones.
Figure 1
(GIF 22.3 KB)
2. Supplementary information on the measurement of psychometric functions for modulation rate
Psychometric functions for modulation rate were measured after completion of each training block using test stimuli with varied modulation rates. For the measurement, the previously used training stimuli were presented (with US reinforcement when necessary) with additional test stimuli of varied modulation rate interspersed between them (without reinforcement) following a standard procedure29. The test stimuli sets were designed to have modulation rates varied in steps of ± 2 octaves/s centered around the modulation rate of the training stimuli in the preceding training block. For example, the training stimuli in training block 1 (rising from 2 kHz to 4 kHz or falling from 4 kHz to 2 kHz) had a modulation rate of 8 kHz/s (2 kHz modulation in 250 ms). Additional test stimuli were therefore designed to obtain a set of stimuli with modulation rates of ± 0, ± 2, ± 4, ± 8, ± 16, and ± 32 kHz/s. As the stimulus duration was held constant at 250 ms, this amounts to frequency modulation ranges of ± 0, ± 0.25, ± 1, ± 2, ± 4, ± 8 kHz, respectively. As further the start frequency was held constant at 2 kHz this particular set of test stimuli consisted of the following sweeps (start frequency [kHz] - stop frequency [kHz ]) : 2-2, 2-2.5, 2-3, 2-4, 2-6, 2-10.
After the second training block a new set of test stimuli was created according to the same rationale and then used to measure psychometric function in addition to the first set of test stimuli. After the third training block, a third set of test stimuli was created and used in addition the previous two test sets, and so on. We also used similar sets of test stimuli with constant duration and end frequency and variable start frequency. Both sets of test stimuli were lumped together in the analysis as they did not produce differing conditioned response (CR) rates.
Before the transition to categorization psychometric functions showed a peak at the modulation rate used in the preceding training block and gradual falloffs of the CR rate similar to the example shown in Fig. 1d in the paper. After the transition to categorization all psychometric functions were found to be of the sigmoid type with high CR rates for all test stimuli with rising modulations (modulation rate 0) and zero CR rate for all test stimuli with negative modulations (modulation rate 0) similar to the example shown in Fig. 1e in the paper. The CR rate was also zero for vanishing modulation (modulation rate = 0), i.e. pure tones, cf. paragraph 1.
3. Testing the clustering of marked states after transition to categorization in the similarity analysis (Fig. 3 in the paper)
The clustering of the spatial patterns after the transition to categorization was statistically analyzed using a resampling approach (Good, P. Permutation Tests. Springer, New York, 2000). A test statistic was defined by
with
and
being the mean distances between patterns across categories and within categories in the categorization phase, respectively; and
and
the corresponding mean values in the discrimination phase. This test value was calculated for the empirical data in the four gerbils and compared with a distribution of T values obtained by randomizing distances between patterns. The distribution was generated by 104 resamples for the data of each animal. The cumulative distribution of T values beyond the empirical T value yielded the attainable signifance level for the one-sided test given as insets to the similarity plots of Fig. 3 in the paper.
Rights and permissions
About this article
Cite this article
Ohl, F., Scheich, H. & Freeman, W. Change in pattern of ongoing cortical activity with auditory category learning. Nature 412, 733–736 (2001). https://doi.org/10.1038/35089076
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.1038/35089076
This article is cited by
-
Modularity in Nervous Systems—a Key to Efficient Adaptivity for Deep Reinforcement Learning
Cognitive Computation (2023)
-
Causal contributions of parietal cortex to perceptual decision-making during stimulus categorization
Nature Neuroscience (2019)
-
Processing complexity increases in superficial layers of human primary auditory cortex
Scientific Reports (2019)
-
A statistical method for analyzing and comparing spatiotemporal cortical activation patterns
Scientific Reports (2018)
-
Go/No-Go task engagement enhances population representation of target stimuli in primary auditory cortex
Nature Communications (2018)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
