At the very first stage of visual processing in the primate, the whole
of visual space is parsed into discrete regions by the neural
circuitry of the retina. By the time signals are prepared to exit the
retina, all that we see travels on about one million fibers from
retinal ganglion cells. At this point, receptive fields (the region of
visual space where a neuron responds to stimuli) are quite small, they
subtend as little as 0.03
in the fovea, and have simple
properties, responding best to small spots of light. Retinal ganglion
cells make synapses onto neurons in the lateral geniculate nucleus of
the thalamus, which in turn project to primary visual cortex (V1).
Here, receptive fields are sometimes still quite small, a fraction of
a degree in some foveal cells. However, as we proceed through
extrastriate cortex and later stages of visual processing, receptive
fields can be quite large, covering entire quadrants of the visual
field.
Our perception of the world, at least on the surface, would appear to conflict with this neural substrate. We can perceive objects as whole even though they almost never are contained by the receptive field of a single neuron. Similarly, a film projector or television actually projects static images in rapid succession, but we readily perceive motion in the projection. Our brain is able to construct a coherent picture of our environment from neurons that take small spatially discrete samples of temporally discrete events.
Within striate and extrastriate visual cortex, a wide range of receptive field properties have been identified. The most famous of these is Hubel and Wiesel's (1962) finding that neurons in primary visual cortex are selective for the orientation of a stimulus (which in their case was a bar of light). More than forty years since their report was published, research has continued to uncover the visual properties of striate and extrastriate cortex. One of the main goals of this work has been to identify how the visual cortex encodes elements of the visual environment that are more complicated than simple bars or edges.
We approached this thesis with two related questions in mind. First, how does the visual cortex piece together the visual scene that spans many receptive fields? Second, how can the responses of cells to simple stimuli in V1 be used to predict their responses and those in higher cortical areas to more complicated stimuli? In Chapters 3 through 7, we will present results from our experiments that attempt to address these questions.
Glass patterns are texture stimuli made by pairing randomly placed dots with partner dots at specific offsets. In Chapter 3, we explore the neuronal signals in V1 that underly the perception of Glass patterns (Glass, 1969; Glass and Perez, 1973). The strong percept of global form that arises from the sparse local orientation cues has made these patterns the subject of psychophysical investigations, yet neuronal responses to Glass patterns have not previously been studied.
In Chapter 5, we report on recordings in area V2, a region of cortex that is thought to be an important early stage in form vision. Here we attempt to build on our experiments and modeling in V1 using Glass patterns. We first extend our findings in V1 to include V2. In addition, we describe the results of an experiment we designed to test the sensitivity of V2 cells to global form cues present outside the classical receptive field.
When two sinusoidal gratings are added together, they form what is called a plaid. In V1, it is possible to reduce the response of a neuron to an optimally oriented grating stimulus by superimposing an orthogonal mask grating. Chapter 6 holds a description of our experiments on the time course of this phenomenon, known as cross-orientation suppression.
Macaque area MT is a cortical region thought to play an important role in motion perception. In this area, plaids have been used to classify cells based on the character of their responses to the pattern. Some neurons (component direction-selective, or CDS) only signal the direction of motion of the component gratings, while others (pattern direction-selective, or PDS) respond to the true direction of the pattern (Rodman and Albright, 1989; Movshon et al., 1985). In Chapter 7, we explain a set of experiments and analysis aimed at using the time course of response to plaid stimuli in MT to learn more about the neural circuitry involved.
In these chapters, we report on experiments designed to address the questions described above. The connections which underly the response properties of neurons in visual cortex are complex and not fully understood. A full description of these properties will require experimentation on and modeling of the spatial and temporal characteristics of neural responses. In our experiments, we have used Glass patterns, gratings and plaids to probe these spatial and temporal response properties in novel ways. It has yielded data that forms another piece of the description of visual cortical neurons. These experiments are further united in their theme of exploring the neural circuitry which combines spatially and temporally discrete signals (i.e., spike trains from neurons) and generates our virtually seamless visual perception of the world.