Visual Perception & Cognition

Overview

The mammalian brain has evolved specifically to cope with changing environments. Confronted with a novel situation, the mammalian brain can plan appropriate actions based not on instinct, but rather on a general cognitive plan – something we call ‘thinking.’

The main goal of our laboratory is to understand, on a cellular level, how the brain represents information, and how single cells and networks can process that information to dynamically perform simple behavioral computations. Many theories exist on how the brain represents and combines information – some suggest that individual cells act as individual storage elements arrayed in hierarchical processing streams, others suggest information is coded in the temporal correlations among sparse, distributed networks of cells. To approach these questions, we must understand:

  1. How information is represented in single cells, i.e. what is the “neural code”
  2. What is physically modified in the brain when something is “learned”, and
  3. How does changing the firing of single units affect perception, or change behavioral output of the brain.

Following are some of the main specific interests of the laboratory:

Changes in intrinsic reflectance during visual stimulation in different parts of the visual field can be used to map retinotopic organization. Shown here is a surface view of the primary visual cortex in squirrel, color-coded for the vertical position of a stimulus in the visual field. Data was obtained by analyzing the phase of the reflectance change to a bar of light cycling from fovea to periphery at 0.1 Hz.

Changes in intrinsic reflectance during visual stimulation in different parts of the visual field can be used to map retinotopic organization. Shown here is a surface view of the primary visual cortex in squirrel, color-coded for the vertical position of a stimulus in the visual field. Data was obtained by analyzing the phase of the reflectance change to a bar of light cycling from fovea to periphery at 0.1 Hz.

Studying the first cortical computations on sensory inputs

Because many elements of cortical architecture are conserved across the whole cerebral cortex, it is reasonable to assume that the basic principles of cortical function apply equally well throughout the cortex. In many mammals, primary visual cortex provides an excellent place to study cortical circuitry due to (1) the ease of providing patterned input (2) the ease of accessibility for recording and imaging methods. Our method is to study patterns of cortical functioning in either normal or perturbed systems by single-unit recording, or imaging methods, in order to learn how these patterns come about and how they can be altered.

The output of a cortical cell can be thought of as a function of the relative balance of excitatory and inhibitory inputs to that cell. Using visual cortex as a model, we showed that pharmacologically blocking the inhibitory inputs to a single cell altered the rate of firing, but did not alter the basic response properties of that cell. By blocking inhibitory activity across many cells in a small area of cortex, we showed that although the activity of those cells substantially increased, the response properties in neighboring areas were only slightly affected. These results argue for a distinction between types of cortical ‘computation’ that are hard-wired, and those that are malleable depending on input activity.

Primary visual cortex is the first level of the visual system where information such as color, orientation, etc. is represented by spatially separated groups of cells. Experiments such as these allow us to examine the importance of local connections and the spatial layout of response properties on visual function.

Comparative functional neuroscience

By studying differences between equivalent circuitry across multiple mammalian species (including humans) we can gain insight into the evolutionary origins of cortical circuitry. Two examples of recent comparative projects are the following:

  1. Cat primary visual cortex has the largest hypercolumn dimensions in any species yet studied. In a graph of spacing verses area of V1, the frequency of iso-orientation domains is roughly half what it should be compared with other animals. Why should this be the case? Is it a clue that some other system of organization is different in the cat?
  2. Mice lack many of the functional structures observed in V1 of other mammals. Mice are rodents that have relatively poor eyesight. Carnivores and primates have well developed spatial organization in V1, but also have excellent eyesight. Is the functional organization of V1 a result of difference in the quality of visual input, or of genetics? To answer this question, we examined the organization of V1 in the squirrel – a rodent with relatively good eyesight. Squirrels were found not to have spatially segregated properties suggesting that it is largely a matter of genetics. Interestingly, this result also suggests that cells do not need to be spatially organized for the animal to have a full repertoire of visual behavior.
Information coding in the brain

An important goal of the brain is to choose appropriate behavioral responses based on changing sensory inputs. We are interested in how the brain chooses a behavioral response. To make this problem tractable, we designed an experiment where subjects had to respond either left or right based on a visual cue that was one of two colors and in one of two locations. In half the trials the cue color indicated which was the correct response to make, and in the other half of trials cue location indicated the correct response to make. Areas of the brain were found that coded for differences in cue color only when color was the behaviorally relevant feature but not when location was the relevant feature. Furthermore, some areas of the brain code specifically for which ‘rule’ is relevant at a given time. These questions can be studied both at a single cell level, and also at a systems level using fMRI. We hope to delineate a behavioral ‘circuit’ that allows us to understand how the brain is able to map instantaneously and arbitrarily between inputs and behavioral outputs. We hope experiments such as these will help describe the processes behind what is commonly referred to as ‘thinking’.

Techniques

Single-unit recording: By recording from single cells using microelectrodes, the outputs of single cells can be recorded, correlated with a given task or input, and analyzed for what information is contained and how it is represented. Cells can subsequently be localized anatomically using tracer injections visualized microscopically in thin sections.

Optical imaging: Using a sensitive camera, we can image either changes in the reflectance of brain tissue that correlates with activity, or changes in fluorescent tracers applied to the brain. These techniques allow us to map various neuronal properties across the surface of cortex. Using fluorescent dyes, we can map anatomical connectivity, patterns of blood flow, or brain activity using dyes that indicate voltage across cell membranes. We can also measure intrinsic signals that allow us to get spatial maps of neuronal activity.

MRI: We are closely affiliated with the Center for Biological Imaging, which houses a 3-Tesla magnet specifically for research use. Both humans and animals can be imaged noninvasively to obtain maps of function, connectivity, chemical content, anatomy, blood flow, and many other modalities. We are interested in pushing the limits of current techniques in MRI for neuroscience.

Publications

Current and forthcoming papers and abstracts from the lab are relevant to the following areas of study: physiology of the visual areas of cerebral cortex, neurophysiology of cognition and comparative neurophysiology, and in functional imaging experiments and techniques.

Kim DS, Ronen I, Olman C, Kim SG, Ugurbil K, Toth, LJ (2003) Spatial relationship between neuronal activity and BOLD functional MRI Neuroimage 21:876-885.

Williams ZM, Elfar JC, Eskandar EN, Toth LJ, Assad JA (2003) Parietal activity and the perceived direction of ambiguous apparent motion Nat. Neurosci. 6:616-623.

Ugurbil K, Toth LJ, Kim DS (2003) How accurate is magnetic resonance imaging of brain function? TINS 26:108-114.

Toth LJ, Assad JA (2002) Dynamic coding of behaviourally relevant stimuli in parietal cortex Nature 415:165-8.

Methodology

Magnetic resonance

Recent advances in MR imaging enable the mapping of brain physiology and anatomy in animals without surgical intervention. As part of the new Biomedical Imaging facility at BUSM, we focus on measuring and improving the spatial accuracy of diffusion tensor imaging (DTI) and functional imaging methods (fMRI), and developing new MR methods useful for neurophysiology. (Collaborating investigators: Dae-Shik Kim, Itamar Ronen.)

Camera-based imaging of neural activity in visual cortex

In experimental preparations, we directly image neural activity using intrinsic signal and voltage-sensitive dye techniques. Most commonly, we use visual stimuli to elicit physiological maps of retinal position, orientation or ocular dominance, and these maps can be used to probe the functioning of cortical circuits under various experimental conditions.

Transcranial magnetic stimulation (TMS)

TMS is a safe way to stimulate human brain tissue non-invasively, and has many potential clinical applications. We use imaging techniques (fMRI, intrinsic signal imaging, and 2DG imaging) to directly visualize neural activity resulting from TMS. The goal is to better understand how to design effective TMS-based clinical treatments. (Collaborating investigators: Antoni Valero-Cabre, Alvaro Pascual-Leone.)

Neurophysiology of behavior in the macaque

Monkeys are excellent animals in which to study the basis of learned behaviors. We are interested in exploring the ways in which primate neurophysiology mirrors human neurophysiology, and in learning about how single cells code information during changing behavioral tasks.

Comparative neurophysiology

Studies of the differences in neuronal physiology between species provide important clues to ask why cortex is organized the way it is. Unique species we are interested in include ferret, squirrel and marmoset.

People

Louis J. Toth, Ph.D.
Dr. Toth is the lab director. He received his doctorate from the Massachusetts Institute of Technology. Dr. Toth is interested in the neurophysiology of cortical computation, the visual system and neuroimaging techniques.

Antoni Valero-Cabre, MD, Ph.D.
Dr. Valero-Cabre conducts experiments on the physiology of TMS using 2DG and optical imaging.

Collaborators from BU Center for Biological Imaging
Dorothe Poggel, Ph.D.
Emi Takahashi Oki, Ph.D.
Fred Powell (MD Ph.D. student)
Sahil Jain (Undergraduate student)