Research (for neuroscientists)
How is sensory information represented by neuronal populations in cerebral cortex? How are sensory representations modified during
learning? Concentrating on the tactile modality, the laboratory is tackling these problems using methods ranging across
electrophysiology, computational modeling, and human psychophysics.
Ancient history
Mathew Diamond's PhD thesis concerned the organization of the cortical body representation in laboratory animals (Diamond, 1989; Favorov and Diamond, 1990; Diamond and Ebner, 1990). Prior to this work, columns in somatosensory cortex had been recognized by Mountcastle based on submodality: neurons responding to a given sort of stimulus (e.g. tactile, joint, muscle) were grouped together in a module and segregated from modules receiving other submodalities. With Favorov and Whitsel, Diamond recognized that within a single submodality the same sort of grouping existed based upon the position of neurons' receptive fields. The cortical body representation, first identified by Penfield, was found to be organized not as a smooth, continuous map but as a mosaic of discrete, but interconnected, columnar modules.
What role does such columnar organization play in the functioning of somatosensory cortex? As a postdoctoral fellow with Prof. Ford Ebner at Brown University, and subsequently as assistant professor at Vanderbilt University, Diamond focused on the phenomenon of cortical map plasticity, which had recently been described by the laboratories of Mike Merzenich and Jon Kaas. Diamond first hypothesized (Diamond and Armstrong-James, 1992), and then demonstrated, that cortex adjusts to changes in sensory experience through modifications in the connections between columnar modules (Diamond et al., 1993; Diamond et al., 1994; also Lebedev et al., 2000). For this work, Diamond was awarded the Cortical Explorer Prize of the Cajal Club.
More recently...
In 1996 Diamond began to set up the SISSA Tactile Perception and Learning Laboratory to continue investigating cortical organization. The main guiding strategy has been to carry out physiological and behavioral experiments in laboratory animals (rats) and psychophysical experiments in humans to look for organizational rules that unify these distant species. In collaboration with Carlo Porro at the University of Udine, an fMRI study found that somatosensory cortex in humans is activated not only during the actual delivery of tactile stimuli, but also during mental imagery of the same stimuli, in the absence of touch (Porro et al., 1996). Fascinated by the finding that somatosensory cortex is not limited to the on-line processing of sensory data, the research projects turned progressively more to the issue of the representation, storage, and recall of information in cortex. Justin Harris (laboratory member, 1999-2002) discovered the principle of "topographic learning": in humans, the neuronal modifications that underlie perceptual learning are distributed according to the spatial framework of the cortical body map (Diamond et al., 2001; Harris et al. 2001a).
Subsequent work extended the principle to short-term memory: the somatosensory cortical map is involved with the storage of information when the stimulus features need to be transiently conserved for subsequent analysis (Harris et al. 2001b, Harris et al., 2002). The somatosensory cortical map thus takes its place as one part of the working memory circuit controlled by prefrontal cortex. Current psychophysical work (Harris et al. 2006) concerns the question of how the perceptual properties of vibration (i.e. frequency, regularity vs noisiness) emerge from cortical activity, and the mechanisms underlying fast perceptual learning - rapid improvements in vibration judgments during training (unpublished).
While human psychophysics is invaluable, problems in neuronal coding and representation cannot be approached directly without experiments in laboratory. The laboratory has been investigating the vibrissal system of rats, selected because the tactile capacities of this sensory system are comparable to those of human finger tip. The lab documented the spatial and temporal distribution of neuronal activity in the tactile region of rat cerebral cortex ("barrel" cortex") using a 100-channel electrophysiological recording system (Rousche et al., 1999; Petersen and Diamond, 2000). Comparing barrel cortex functional organization to the distribution of tactile learning, Diamond's laboratory showed that the "topographic learning" principle generalizes from humans to rats - tactile information is stored within the spatial structure of maps (Harris et al., 1999; Diamond et al., 1999). In a related set of experiments with Rasmus Petersen and Stefano Panzeri, the laboratory used rigorous Information Theory methods to investigate exactly which features of ensemble neuronal activity report information about whisker stimuli (e.g. Panzeri et al., 2001; Petersen et al., 2001; Petersen et al., 2002).
Now
Recently, the laboratory has been developing a quantitative understanding of how neurons at multiple levels of the sensory pathway encode dynamic, complex whisker stimuli in their spike trains. This involves analyses of neuronal activity during computer-controlled "noisy" stimuli (Lak et al., 2008) as well as with natural textural stimuli. We have monitored the vibrations produced in the shaft as the whisker sweeps across different surfaces and have identified distinctive patterns of neuronal activity induced by such vibrations. These patterns seem to be the neuronal "signature" for textures (Arabzadeh et al., 2005;Hipp et al., 2006;Arabzadeh et al., 2006).
Having generated precise hypotheses from studies of anesthetized animals, we set out to study the nature of the cortical representations underlying judgments of texture in awake, behaving rats (von Heimendahl et al., 2007). In the behaving animal, the whisker sensory system is ‘‘active’’: the animal generates sensory signals by palpating objects through self-controlled whisker motion (just as we move our fingertips along surfaces to measure their tactile features). Active sensing (Kleinfeld et al.,2006;Diamond et al.,2008a)endows the animals with the capacity for fast and accurate texture discrimination (although, from the experimenter’s point of view, the animal's own control over whisking is an enormous complication!). We recorded neuronal activity from barrel cortex while rats used their whiskers to discriminate between rough and smooth textures. On whisker contact with either texture, firing rate increased by a factor of two to ten. Average firing rate was significantly higher for rough than for smooth textures, and we therefore propose firing rate as the fundamental coding mechanism. The rat, however, cannot take an average across trials, but must make an immediate decision using the signals generated on each trial. To estimate single-trial signals, we calculated the mutual information between stimulus and firing rate in the time window leading to the rat’s observed choice. Activity during the last 75 ms before choice transmitted the most informative signal; in this window, neuronal clusters carried, on average, 0.03 bits of information about the stimulus on trials in which the rat’s behavioral response was correct. To understand how cortical activity guides behavior, we examined responses in incorrect trials and found that, in contrast to correct trials, neuronal firing rate was higher for smooth than for rough textures. Analysis of high-speed films suggested that the inappropriate signal on incorrect trials was due, at least in part, to nonoptimal whisker contact. In conclusion, this experiment suggests that barrel cortex firing rate on each trial leads directly to the animal’s judgment of texture, as discussed in Diamond et al.,2008a, b.
This work provides us with many new insights about how to "decode" what the rat has touched only by seeing its neuronal activity. This, after all, is exactly what the rat's brain must accomplish! We are now particularly interested in higher-level representations of tactile stimuli – how the physical parameters of whisker motion become transformed to a representation of object identity.
last modified: 20 October 2011