The sense of vision seems effortless, and intuitively visual perception seems like it should be simple. However, research has shown that visual scene analysis requires, in addition to the retinas and numerous brainstem nuclei, the use of about one-third of the cerebral cortex. This is because visual perception (as opposed to the mere image recording seen in a digital camera) means taking the pattern of light and dark from a 2D image and from it inferring the structure and properties of objects in the real 3D world. Vision is not a low-level process, but a higher cognitive process whose study has drawn the attention of a large fraction of the neuroscientists. The rhesus monkey is the best available model of the human visual system. This project will involve recording from single V4 cortical neurons from behaving rhesus monkeys to multiple simultaneous visual stimuli. Visual cortical area V4 is a major cortical center for form perception. Unlike so-called "higher" cortical areas such as the inferotemporal (IT) cortex, neurons in V4 can be easily stimulated with simple geometric shapes. Unlike so-called "lower" areas such as V1 and V2, neurons in V4 have receptive fields large enough to encompass more than one discrete visual stimulus at a time. These properties mean that V4 is an ideal location to study how neurons in the visual system combine information from different parts of an image. Neurons early in the visual system examine only a small part of an image (i.e., they have small receptive fields), but neurons in later areas of processing have much larger receptive fields. Neurons with large receptive fields cannot be simply summing their inputs, or the result would be objects that appear as just a large blur. Neurons must be processing their inputs in a manner other than averaging or summing. But what are the rules of this processing? Experimental and theoretical studies suggest that one set of rules is for a neuron to select only the inputs that by themselves would elicit the maximum response. A competing theory is that neurons should compute the weighted average of their inputs. This study aims to resolve this controversy, an aim that is of significant importance to the field of visual neuroscience, and potentially, of importance to computer vision. While the focus of this project is on vision, in principal the results could apply to all of the cerebral cortex. Nature is conservative, and rarely creates a mechanism that is only used on just one location. Also, cerebral cortex is notable for it's relative homogeneity, i.e, the neural circuits in part of the cerebral cortex follow generally the same organization as the circuits on another place in cortex. Finally, the cerebral cortex is notable for the richness of reciprocal connections between different regions. Indeed, it is this richness of interconnections that makes the cerebral cortex so susceptible to seizures. A core part of how the cerebral cortex functions must be how a given region combines or selects from the richness of inputs available to it. As such, this study aims to explore not just the visual system but fundamental aspects of cortical function.
The broader impact of the research is to promote interdisciplinary teaching and training of both undergraduate and graduate students. There will be a focus on encouraging participation from members of underrepresented groups from several minority serving institutions in Alabama.
If the world only ever contained a single simple shape, vision would be an easily understood process. However, the world contains many objects that are often made of many parts and whose arrangement in space can be complex. How the visual parts of the brain deal with this complexity, to give us what appears to be an effortless ability to tell what is where just by looking, remains imperfectly understood. This project explored the rules by which neurons in the brain combine signals from multiple objects and multiple pieces of single objects. There were several interesting results, but the primary one was realizing that short time-scale dynamics are important for this process. In particular, it was demonstrated that temporal gating – where early-arriving inputs can block later-arriving inputs – is a powerful effect in the cerebral cortex of awake behaving primates. The figure shows example responses from a single primate visual cortical neuron that exhibited strong temporal gating effects (redrawn from Gawne, T. (2013) Cortical Computations Using Relative Spike Timing. In, Spike Timing: Mechanisms and Function. Eds: Patricia Di Lorenzo and Jonathan Victor. Taylor & Francis. ISBN 13:978-1-4398-3815-0). Panel A. Response to a single high-contrast stimulus presented alone. The continuous heavy line is the average firing rate as a function of time with the standard error of the mean (an indicator of the reliability of the data) indicated by the thin lines to either side. The rasters (time of single ‘spike’ occurrences, where a spike is the digital code that a neuron uses to communicate long distances with other neurons) are shown in light gray. Stimulus configurations within the receptive field (the location in visual space that a neuron responds to) of the neuron are illustrated in the dashed ellipses. Panel B. Response to a single stimulus at another location. Panel C. Response to both stimuli presented at the same time. Panel D. All responses overlaid. The response to the high contrast stimulus presented alone has a shorter latency than the response to the low contrast stimulus presented alone. The response to both stimuli presented at the same time tracks the short-latency response with little effect from adding the stimulus that, by itself, elicits a longer latency response. In other words, the first arriving stimulus locks-out the later arriving stimulus. A variety of other novel findings sprang from the seemingly simple procedure of recording the neuronal responses to multiple visual stimuli, including: a new approach to seeing how the effects of selective attention affect the operation of the brain on a short time-scale, new approaches to analyzing electrical fields in the brain, including new approached to analyzing the human electro-encephalogram (EEG) which is currently being evaluated for possible clinical applications, new insights into how the visual system processes blur (a sharp edge may be considered to consist of the sum of a blurry edge and the high-spatial frequency components that are the difference between a sharp and a blurry edge). This grant is now completed, but the results continue to spur new developments in a variety of areas in what is termed systems neuroscience (the neuroscience that is concerned not with molecules and genes, but the computations and information-processing performed by the brain). The main conclusion, however, is this: neurons are not simple devices that just count the number of action potentials, but they appear to perform careful and precise computations on their inputs using the relative timing of spikes from different inputs at the millisecond level of resolution.
