In one study, we are exploring several methods to map the brains functional MRI responses to social movies in a meaningful way. Human studies suggest that evaluating the reliability of fMRI signals in individual voxels to multiple presentations of the same movie provides a fresh alternative to conventional block or event-related design. This data-driven approach, which we term repeated viewing correlation (RVC), circumvents the need for a priori modeling of responses and allows for the study of processes involving complex or unpredictable response dynamics. At the same time, the RVC approach alone cannot disentangle the multiple factors that will contribute to a voxels time course, since voxels with similar RVC values can be engaged by very different features. Here we combine the RVC approach with a model-based approach in which maps are constructed based on coefficients derived from multiple linear regression. Stimulus models were obtained both by subjectively scoring the videos visual and social content (i.e. an ethogram) and by objectively analyzing the time-varying image-based statistics of the movie. We collected fMRI data from three monkeys, who watched 18 video-only short movies varying in social content. In total, we collected over 700 five-minute whole brain fMRI scans during periods of free viewing in which the eye position was monitored. The RVC approach revealed a prototypical activation map that was common across all the movies, along with certain regions whose RVC values depended upon the specific content. The Pearsons correlation coefficient typically exceeded 0.5 in the retinotopic visual cortex, the inferotemporal cortex, the lateral intraparietal cortex, and the ventral premotor cortex. This is in contrast to many other regions, such as in the white matter and auditory cortex, which exhibited correlation values at, or close to, 0.0. Though the RVC value was high among active regions, their time courses often differed markedly. We therefore examined the basis of these differences using the various stimulus models we obtained from the scoring of the videos. Regression coefficients provided robust mapping of brain regions sensitive to low-level features (e.g. speed) from those selective to social stimuli (e.g. faces) and behaviors (e.g. aggression). Taken together, these approaches provide multiple lenses onto the brains responses to dynamic scenes and contexts that cannot easily be incorporated into a conventional fMRI design. In another study, we are expanding this approach by applying the RVC approach to single-unit activity in the inferior temporal cortex. Neurons in the inferior temporal (IT) cortex respond to diverse types of complex stimuli, including natural categories such as faces or biological movement. Monkey electrophysiology experiments typically assess neural response selectivity by presenting multiple, short duration stimuli while the animal fixates a small point. The responses of a given neuron are typically evaluated over the course of an hour or two. Here we consider how IT neurons respond under more naturalistic viewing conditions, in which the animal freely views dynamic videos of conspecifics and heterospecifics in a range of social contexts. Using chronically implanted microwire arrays (Bondar et al, 2009), we further evaluate the extent to which neurons respond similarly to a video over much longer periods. Three rhesus macaques repeatedly viewed up to 12 different five-minute movies varying in social content. We collected spiking and broadband field potential responses from 32 and 64 channel microwire bundles chronically implanted in the lower bank and fundus of the superior temporal sulcus (STS). Analysis of gaze position revealed a high degree of overlap in fixation targets between viewings, and particularly those surrounding salient social events. Moreover, epochs of high and low spiking were strikingly consistent between multiple viewings of the same movie, even when they were not obviously linked to specific stimulus variables such as faces, bodies, or movement (Fig. 1). Analysis of single units maintained over multiple sessions (up to four months) revealed that such consistency was also observed across different days and even over several weeks. Neural responses to individual events were robust to shuffling of the individual scenes at approximately 40 second long time scales. These results demonstrate that the repeated viewing approach provides a window into the role of individual neurons in the encoding of complex stimulus sequences, even under conditions in which the primary determinants of neural responsiveness cannot be summarized in terms identifiable high- or low-level stimulus features. Finally, we are using the microarrays to examine the long-term plasticity of IT selectivity by longitudinally tracking response patterns. Neural plasticity in high-level visual cortex is believed to mediate our ability to acquire new perceptual expertise. According to this idea, experience drives changes in how neurons in the brain process visual inputs, and these neuronal changes in turn lead to the behavioral signs of learning. Evidence supporting this view comes primarily from acute physiological recordings from single neurons in inferotemporal (IT) cortex that compared the visual responses to trained vs. untrained stimuli. Because conventional microelectrode methods can only follow activity from single neurons in an awake monkey for one or two hours at the most, previous experiments in behaving monkeys necessarily involved comparing """"""""snapshots"""""""" of neuronal responses before and after training. Thus it remains unclear whether plasticity in inferotemporal cortex is a cause or a consequence of visual learning. To understand whether and how long-term learning and plasticity in IT cortex are causally related, we ultimately need to conduct longitudinal electrophysiological experiments that track visual responses across days as the animals acquire new perceptual expertise. To this end, we developed a chronic recording array of inertialess microwires that were capable of following the activity of single neurons across multiple days. We first used this technique to demonstrate that, in the absence of explicit learning pressure, the pattern of visual responses evoked by large sets of stimuli is remarkably stable over periods of as long as one month. We then asked how the acquisition of new perceptual expertise affects the visual responses of IT neurons. Monkeys were trained to sort large stimulus sets into two different categories based on reward outcome. Learning altered the visual responses in 68% (21/31) of longitudinally recorded spiking responses. These changes included both activity increases and decreases, and in some cases lead to the genesis of new spiking responses to stimuli that were previously ineffective at driving the cell. Learning-related changes were also evident in the local field potential. Although the monkeys'behavior indicated that perceptual learning was essentially complete after a one hour training session, the learning-induced changes in IT visual responses did not emerge until 24 hours later. This delayed expression of neural plasticity indicates that changes in IT visual responses are a consequence of learning, rather than the causal driver of learning, and may reflect a process of memory consolidation.

Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
U.S. National Institute of Mental Health
Zip Code
Russ, Brian E; Kaneko, Takaaki; Saleem, Kadharbatcha S et al. (2016) Distinct fMRI Responses to Self-Induced versus Stimulus Motion during Free Viewing in the Macaque. J Neurosci 36:9580-9
Miller, Cory T; Freiwald, Winrich A; Leopold, David A et al. (2016) Marmosets: A Neuroscientific Model of Human Social Behavior. Neuron 90:219-33
Murphy, Aidan P; Leopold, David A; Humphreys, Glyn W et al. (2016) Lesions to right posterior parietal cortex impair visual depth perception from disparity but not motion cues. Philos Trans R Soc Lond B Biol Sci 371:
Hung, ChiaChun; Day-Cooney, Julian; Russ, Brian et al. (2015) Neural responses to naturalistic movies in the common marmoset using electrocorticography and fMRI. J Vis 15:580
McMahon, David B T; Russ, Brian E; Elnaiem, Heba D et al. (2015) Single-unit activity during natural vision: diversity, consistency, and spatial sensitivity among AF face patch neurons. J Neurosci 35:5537-48
Hung, Chia-Chun; Yen, Cecil C; Ciuchta, Jennifer L et al. (2015) Functional mapping of face-selective regions in the extrastriate visual cortex of the marmoset. J Neurosci 35:1160-72
Dougherty, Kacie; Cox, Michele A; Ninomiya, Taihei et al. (2015) Ongoing Alpha Activity in V1 Regulates Visually Driven Spiking Responses. Cereb Cortex :
Russ, Brian E; Leopold, David A (2015) Functional MRI mapping of dynamic visual features during natural viewing in the macaque. Neuroimage 109:84-94
Mitchell, Jude F; Leopold, David A (2015) The marmoset monkey as a model for visual neuroscience. Neurosci Res 93:20-46
Izpisua Belmonte, Juan Carlos; Callaway, Edward M; Caddick, Sarah J et al. (2015) Brains, genes, and primates. Neuron 86:617-31

Showing the most recent 10 out of 18 publications