Synopsis

Research in this laboratory concentrates on the neural mechanisms of perception and object recognition. Although our basic research revolves around vision, a number of independent collaborators are also investigating the relationship between neural activity and perception using other sensory modalities. I firmly believe that such scientific questions require a multimodal methodological approach that integrates information obtained from single units with that derived from mass action potentials as well as from a number of activity-related, surrogate signals such as those monitored during noninvasive neuroimaging experiments. Parallel to our ongoing neuroscientific research, therefore, we are also working to develop methodologies that will permit us to the study neural networks in the context of behavioral paradigms. We have already designed and implemented two high-field magnetic resonance imaging systems for functional, anatomical and spectroscopic imaging. The systems are endowed with all the necessary hard and software to conduct simultaneous imaging and recordings, and they are being used to study the function, connectivity, and neurochemistry of the non-human primate brain. Furthermore, while continuing to exploit traditional neuroimaging in our experiments, we are also investigating the relationship of neural activity to the MR-measurable hemodynamic responses and experimenting with methods that do not rely on hemodynamic responses at all. In the context of the last-named project, a group of synthetic and coordination chemists in my laboratory are attempting to synthesize and evaluate MR-detectable smart probes that change magnetic properties as a function of the concentration of ions and molecules involved in neural signaling.  Smart contrast agents, if designed and tested appropriately, promise to revolutionize invasive neuroimaging and would represent a quantum leap forward in signal-to-noise ratio, spatial detail and specificity, while affording unprecedented temporal resolution.

Neurons, Networks and Perception

There is often more than one way of 'seeing' something, of perceiving, recognizing and understanding it. The adjacent twin images below are best viewed through a stereoscope, but experts need no viewers. In fact, you are probably an expert if you have bothered to read this far.

The images can be fused by simply going cross-eyed until the two pictures are superimposed. Wait for a moment until the brain adjusts the eye-focus to a distance longer than that predicted by the strong convergence of your eyes. Suddenly, you will clearly see a 3-dimensional image of a room full of objects showing perspective reversals, ambiguous patterns, and labile figure-ground organizations. Each can be seen in more than one way. Look at the little section of staircase, for example, or the "missing cube" figure that is seen one moment as a block with a cube missing from the corner nearest to the viewer, and the next moment as a cube hanging from the upper corner of a room. Both objects change perspective, even though the stereoscopic information disambiguates their three-dimensionality.

Why? What makes perception unstable - or stable, for that matter? (At this point it bears recalling that such perceptual multi-stability affects other senses as well. We just happen to be working with the visual system, so the examples that we present here will concentrate on visual phenomena.) So, what determines what we see at any given time? We know that a very large number of visually driven neurons are activated by any such given pattern. And perhaps it is even the same neurons that are activated for multiple interpretations. Is the perceptual shift due to small changes in the activity of each one of these neurons? Is the awareness of the stimulus the result of activation of cells in specific visual areas, or is it the outcome of neural activity underlying specific image-analysis stages, and thus unrelated to anatomical or functional area boundaries? And what determines the “shift”? For several years my collaborators and I have turned our energies to detecting and studying the neural correlates of these perceptual alternations.

So, you might ask, are all of these illusions and puzzles good for anything beyond, say, livening up a rather boring party? Our answer is a resounding "Yes!" All of perception is itself an illusion. Our brain must constantly filter the jumble of signals that bombard us and, based on prior experience, make educated guesses at what we are seeing. In effect, all it can do is simply decide what would make the most sense. It is precisely the cases where, this “best guess” method breaks down, that can tell us the most about how perception actually works.

In our lab, we have been experimenting with so-called binocular rivalry. For an example, look at the Caneja patterns at right – fuse the upper pair, then the lower. Do you see any difference? Rivalry is a fascinating bistable phenomenon and optimally suited for experiments with animals. A number of ingenious psychophysical experiments by Willem Levelt, Randy Blake, Robert Fox and others revealed so much about the properties of this phenomenon and its putative neural basis that it became clear that we could no longer afford to overlook physiology.

Accordingly we trained monkeys to report perceptual alternations and recorded from many different visual cortical areas. The results from these studies indicated that the activity of only a fraction of the brain’s neurons seems to be related with what the animals perceive at any given time. Notably, even though perception-responsive cells are concentrated in cortical areas near the top of the processing hierarchy, they can be found all along the visual pathway. Evidently (and in my view not suprisingly) a highly interconnected neural network of these cells determines the awareness of a stimulus (see reviews listed in the end of the page). As exciting as this is, it makes the task at hand seem even more daunting. How can we ever grasp the inward, hidden nature of the workings of such a network (or networks)?

I believe that our only hope of addressing such questions will require us to abandon our obsession with single neurons and take the development and application of integrative approaches seriously. Single cell recordings, as important and unique as the information that they provide us with may be, fall short of affording us insight into the organization of networks. Large electrode- or tetrode-array recordings, monitoring of action potentials and slow waves, in vivo connectivity studies, and neuroimaging must all be employed to obtain the kind of information required for studying the brain’s capacity to generate various complicated behaviors. The continuous development and improvement of anatomical, physiological and imaging techniques has been - and will continue to be - instrumental in bringing about the integration of such information.

Over the last 7 years our laboratory has been intensively engaged in the development of the magnetic resonance imaging technique in the non-human primate and its combination with different traditional neuroscientific methods, including tracer studies for connectivity, intracortical recordings, microstimulation, and investigations of neurochemistry.

Neuroimaging and Intracortical Recordings

This combination of methodologies currently allows the study of (a) the long-range (lateral and feedback) interactions between different brain structures, (b) task- and learning-related neurochemical changes by means of localized in vivo spectroscopy or MRS, (c) dynamic connectivity patterns by means of labeling techniques involving MR contrast agents, and (d) plasticity and reorganization following experimentally placed focal lesions. Such applications are likely to bridge the gap between human neuroimaging studies and the large body of animal research carried out over the last half century.

We have been using two high-field magnetic resonance imaging systems for functional, anatomical and spectroscopic imaging (the left figure shows the 4.7T and 7T scanners at the left and right panel respectively). The systems are endowed with all the necessary hard and software to conduct simultaneous imaging and recordings. High resolution MR Imaging (see right image) with these system can reveal anatomical detail usually obtained in histological preparations.

Among other things, our combined physiology and imaging studies also provided the first data on the nature of the signals measured in functional magnetic resonance imaging (fMRI) experiments. In a first approximation, BOLD responses and neural responses are shown to have a linear relationship for stimulus presentations of short duration. The hemodynamic response appears to be better correlated with the local field potentials, implying that activation in any given area is often likely to reflect the incoming signals and the local processing in that area rather than the spiking activity. Current work is investigating the neural activity changes observed in brain areas exhibiting stimulus anti-correlated responses, a phenomenon often referred to as “negative BOLD.” At the same time, physiology and fMRI experiments are in progress to examine the correlation between different neural signals (local field potentials and multiple unit activity, for example) and perfusion, as seen in cerebral blood volume changes and the cerebral oxygen metabolism rate.

Finally, a team of chemists in my laboratory is working to develop smart contrast agents (SCAs). Current neuroimaging relies on hemodynamics and would be substantially improved if we could carry it out in conjunction with SCAs. SCA technology exploits the configuration changes occurring in complexated ligands in the presence of other ions or molecules such as calcium, potassium or certain signaling molecules whose concentration changes parallel the time course of neural activation. Their successful application in neuroscience is likely to usher in a real revolution, as it promises truly spectacular spatiotemporal resolution and specificity for whole-brain imaging.