The brain is exquisitely sensitive to the transitional statistics of environmental events. It uses these statistical relationships to form predictions about future outcomes. These predictions form the basis for a broad range of cognitive abilities, including infant language acquisition(1), feature invariance(2), and visual object recognition(3). While statistical learning has been extensively studied in humans, we lack a comprehensive understanding of its neural circuitry due to the need for an appropriate animal model. I have recently developed a model to study prediction coding at the single neuron level in non-human primates of statistically learned visual stimuli. Over the course of my training I’ve used single unit, multi-electrode, local field potentials, and human electroencephalogram (EEG) recordings. My principal current research goal is to use cutting-edge chronic recording techniques in awake monkeys to develop a mechanistic understanding of the neural circuitry of statistical learning, and its relation to other learning mechanisms.
Previous Research - Distribution of Working Memory in the Prefrontal Cortex
For my dissertation project, I joined Prof. Christos Constantinidis’ laboratory at Wake Forest Medical Center, with the goal to better understand how learning influences the prefrontal cortex. At the time there was disagreement regarding the functional organization of the prefrontal cortex (PFC). Earlier studies based on anatomical, lesion, and physiology research proposed that the lateral PFC was systematically organized: spatial working memory was represented in the dorsal PFC and shape/feature working memory was represented in the ventral PFC(4), by analogy to the what/where paths in afferent visual regions. Later studies suggested however, that what/where organization could be reshaped if the training required remembering the combination of shape plus location (conjunctions) of the stimuli (5). My research revealed that even before training on any working memory tasks, neurons across the lateral PFC were selective for spatial, shape, and conjunction features (6). The distribution of spatial and shape selectivity conformed to a what/where pattern and was unaffected by training on spatial, shape, or conjunction features (Fig 1). I’ve described these findings in a number of journal publications (6-11). My graduate work was funded by a predoctoral NRSA.
Current Research - Prediction Coding: Understanding Vision through Learning
For my postdoctoral research, I joined Prof. Carl Olson’s laboratory at Carnegie Mellon University in order to study learning in the inferior temporal cortex (IT). Early on I realized that while there is a wealth of research on statistical learning in infants and adult humans, there is little understanding of the underlying neural mechanisms. I first sought to develop an animal model of statistical learning in the visual system of non-human primates to afford the use of extra-cellular recordings.
The brain presumably handles the vast computational challenge of visual information processing by making predictions of future objects (12). If the brain is accurate in its prediction, then the need to fully process the image is reduced. My research revealed that repeatedly exposing monkeys to fixed sequences of images over the course of weeks, led to underlying neural changes in inferotemporal cortex neurons. Once the leading image in the sequence has become a reliable predictor for the second image, the neural response to the trailing image, when presented in the trained context, is reduced (Fig 2). We termed this phenomenon ‘prediction suppression’ (13). We have completed numerous additional studies characterizing this phenomenon, some of which are in print (14), and others are in review.
A similar form of prediction coding has been theorized for the processing of single images (15). Through experience, objects become more familiar leading to reduced response strength and more efficient coding. To test this idea we created an experimental paradigm using rapid serial visual presentations of objects that compared alternating familiar images to alternating novel images. We found that IT neural responses to familiar images were truncated but that this conferred an advantage in allowing neurons to return to baseline and respond to the next image (Fig 3.). I collaborated with researchers at the University of Pittsburgh Psychiatry Department to search for an equivalent effect in human subjects using EEG recordings. We found an effect in humans analogous to the effect in monkeys. This study was published in Nature Neuroscience(16). Portions of my postdoctoral work were funded by a postdoctoral NRSA.
1. Saffran JR, Aslin RN, Newport EL. Statistical learning by 8-month-old infants. Science. 1996;274(5294):1926-8. Epub 1996/12/13. PubMed PMID: 8943209.
2. Li N, DiCarlo JJ. Unsupervised natural experience rapidly alters invariant object representation in visual cortex. Science. 2008;321(5895):1502-7. Epub 2008/09/13. doi: 10.1126/science.1160028. PubMed PMID: 18787171; PubMed Central PMCID: PMC3307055.
3. Bruner JS. On perceptual readiness. Psychological review. 1957;64(2):123-52. Epub 1957/03/01. PubMed PMID: 13420288.
4. Goldman-Rakic PS. Regional and cellular fractionation of working memory. Proceedings of the National Academy of Sciences of the United States of America. 1996;93(24):13473-80. PubMed PMID: 8942959; PubMed Central PMCID: PMC33633.
5. Rao SC, Rainer G, Miller EK. Integration of what and where in the primate prefrontal cortex. Science. 1997;276(5313):821-4. PubMed PMID: 9115211.
6. Meyer T, Qi XL, Constantinidis C. Persistent discharges in the prefrontal cortex of monkeys naive to working memory tasks. Cereb Cortex. 2007;17 Suppl 1:i70-6. Epub 2007/11/21. doi: 10.1093/cercor/bhm063. PubMed PMID: 17726005.
7. Meyer T, Qi XL, Stanford TR, Constantinidis C. Stimulus selectivity in dorsal and ventral prefrontal cortex after training in working memory tasks. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2011;31(17):6266-76. Epub 2011/04/29. doi: 10.1523/JNEUROSCI.6798-10.2011. PubMed PMID: 21525266; PubMed Central PMCID: PMC3103869.
8. Qi XL, Meyer T, Stanford TR, Constantinidis C. Changes in prefrontal neuronal activity after learning to perform a spatial working memory task. Cereb Cortex. 2011;21(12):2722-32. Epub 2011/04/30. doi: 10.1093/cercor/bhr058. PubMed PMID: 21527786; PubMed Central PMCID: PMC3209796.
9. Qi XL, Meyer T, Stanford TR, Constantinidis C. Neural correlates of a decision variable before learning to perform a match/non-match task. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2012;32(18):6161-9. Epub 2012/05/04. doi: 10.1523/JNEUROSCI.6365-11.2012. PubMed PMID: 22553022; PubMed Central PMCID: PMC3350750.
10. Katsuki F, Qi XL, Meyer T, Kostelic PM, Salinas E, Constantinidis C. Differences in Intrinsic Functional Organization Between Dorsolateral Prefrontal and Posterior Parietal Cortex. Cereb Cortex. 2013. Epub 2013/04/03. doi: 10.1093/cercor/bht087. PubMed PMID: 23547137.
11. Qi XL, Katsuki F, Meyer T, Rawley JB, Zhou X, Douglas KL, Constantinidis C. Comparison of neural activity related to working memory in primate dorsolateral prefrontal and posterior parietal cortex. Frontiers in systems neuroscience. 2010;4:12. Epub 2010/06/02. doi: 10.3389/fnsys.2010.00012. PubMed PMID: 20514341; PubMed Central PMCID: PMC2876875.
12. Rao RP, Ballard DH. Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. Nature neuroscience. 1999;2(1):79-87. Epub 1999/04/09. doi: 10.1038/4580. PubMed PMID: 10195184.
13. Meyer T, Olson CR. Statistical learning of visual transitions in monkey inferotemporal cortex. Proceedings of the National Academy of Sciences of the United States of America. 2011;108(48):19401-6. Epub 2011/11/16. doi: 10.1073/pnas.1112895108. PubMed PMID: 22084090; PubMed Central PMCID: PMC3228439.
14. Meyer T, Ramachandran S, Olson CR. Statistical learning of serial visual transitions by neurons in monkey inferotemporal cortex. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2014;34(28):9332-7. doi: 10.1523/JNEUROSCI.1215-14.2014. PubMed PMID: 25009266; PubMed Central PMCID: PMC4087210.
15. Mumford D. On the computational architecture of the neocortex. II. The role of cortico-cortical loops. Biological cybernetics. 1992;66(3):241-51. Epub 1992/01/01. PubMed PMID: 1540675.
16. Meyer T, Walker C, Cho RY, Olson CR. Image familiarization sharpens response dynamics of neurons in inferotemporal cortex. Nature neuroscience. 2014. doi: 10.1038/nn.3794. PubMed PMID: 25151263.
17. McMahon DB, Bondar IV, Afuwape OA, Ide DC, Leopold DA. One month in the life of a neuron: longitudinal single-unit electrophysiology in the monkey visual system. Journal of neurophysiology. 2014;112(7):1748-62. doi: 10.1152/jn.00052.2014. PubMed PMID: 24966298; PubMed Central PMCID: PMC4157170.
18. McMahon DB, Jones AP, Bondar IV, Leopold DA. Face-selective neurons maintain consistent visual responses across months. Proceedings of the National Academy of Sciences of the United States of America. 2014;111(22):8251-6. doi: 10.1073/pnas.1318331111. PubMed PMID: 24799679; PubMed Central PMCID: PMC4050553.
19. McKee JL, Riesenhuber M, Miller EK, Freedman DJ. Task dependence of visual and category representations in prefrontal and inferior temporal cortices. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2014;34(48):16065-75. doi: 10.1523/JNEUROSCI.1660-14.2014. PubMed PMID: 25429147; PubMed Central PMCID: PMC4244472.
20. Nieder A, Freedman DJ, Miller EK. Representation of the quantity of visual items in the primate prefrontal cortex. Science. 2002;297(5587):1708-11. doi: 10.1126/science.1072493. PubMed PMID: 12215649.