During all electrophysiological experiments reported here, we exclusively targeted the middle temporal area (MT), also known as V5. MT is common to all primates (Kaas and Lyon, 2001), which is one of many reasons why MT has been the area of choice in numerous studies investigating the influences of cognitive processes on sensory information processing in single neurons. In the following section the properties of area MT most relevant to the present studies will be reviewed.
1.2.1 Location and structure
MT is part of the dorsal pathway and is located in the posterior bank of the superior temporal sulcus (STS). Like V1, MT is retinotopically organized (Van Essen et al., 1981), with each hemisphere containing a complete map of the contralateral visual hemifield and a small portion of the ipsilateral field close to the vertical meridian (Desimone and Ungerleider, 1986). Foveal vision is markedly emphasized, with the central 15˚ of the visual field occupying over half of MT’s surface area (Van Essen et al., 1981). There is also a biased representation of the lower visual field quadrant.
Foveal vision and the lower visual field are represented in the lateral part of MT, while larger eccentricities and the upper visual field are represented more medially (Maunsell and Van Essen, 1987). The vast majority of neurons in area MT show strong selectivity for processing of motion direction and speed (see also ‘Response properties of MT neurons’, pp. 8). Neurons sharing similar direction preferences are clustered in columns oriented perpendicular to the cortical surface (Albright et al., 1984). All directions are uniformly represented in MT neurons with motion direction preference changing gradually in adjacent columns.
1.2.2 Connectivity
MT represents an intermediate stage within the hierarchy of visual information processing. It receives feedforward inputs from multiple cortical areas, including V1, V2, V3, V3A, VP, and PIP (Felleman and Van Essen, 1991; Maunsell and van Essen, 1983). The main input to MT, however, is a mainly magnocellular (M) projection, originating from direction and speed selective complex cells in V1 (Movshon and Newsome, 1996). Even though the cortical inputs to MT predominate, some MT neurons remain both visually responsive and even direction-selective after removal or
1.2 Middle temporal visual area (MT) 7 inactivation of V1. This residual functionality might derive from callosal connections from the intact hemisphere (Girard et al., 1992), or direct subcortical inputs from the SC (Rodman et al., 1990) and koniocellular neurons of the LGN (Nassi and Callaway, 2006; Sincich et al., 2004). Nevertheless, MT and the dorsal stream rely heavily on visual information provided by the magnocellular pathway. This is demonstrated by the fact that reversible inactivation of the M layers of the LGN nearly completely abolishes the visual responsiveness of MT neurons, whereas P-layer inactivation has a much smaller, though measurable, effect (Maunsell et al., 1990).
The main cortical target regions for feedforward projections arising in MT are its neighboring areas FST and MST in the STS, parietal lobe areas such as VIP, LIP and 7a and also frontal lobe areas such as FEF and the dorsolateral prefrontal cortex.
In addition, MT also has multiple feedback projections to cortical (V1, V2, V3A) and numerous subcortical (e.g. dorsal LGN, pulvinar, and SC) regions. For example, MT inactivation affects orientation and direction selectivity in V2 neurons, indicating that MT feedback projections influence these neurons’ RF properties (Gattass et al., 2005).
1.2.3 Receptive field structure of MT neurons
The classical RF diameters of MT neurons are about equal to their eccentricity and therefore about 10 times larger than the diameter of their V1 inputs (Born and Bradley, 2005). As a consequence, considerable spatial pooling arises in the formation of MT cells’ RFs. About half of the neurons in MT have RFs with direction selective antagonistic surrounds, which on average spread across an area about three times the size of the classical RF diameter (Allman et al., 1985; Raiguel et al., 1995; Tanaka et al., 1986). A stimulus extending outside the classical RF of an MT neuron with antagonistic surround suppresses the neurons response. This effect is strongest, when the stimulus motion in the surround represents the neurons preferred direction.
Surround suppression is contrast dependent and vanishes at low contrast (Pack et al., 2005). MT neurons will therefore respond better to a large stimulus with low contrast than to one with high contrast. In the macaque, neurons with antagonistic surround RFs are more common in the output layers, whereas those lacking antagonistic surrounds are found predominantly in the input layers (Raiguel et al., 1995).
8 Chapter 1. Introduction 1.2.4 Response properties of MT neurons
Like other areas of the superior temporal sulcus (MST, FST) MT contains many cells sensitive to the direction of motion and the speed of a stimulus (Dubner and Zeki, 1971; Maunsell and Van Essen, 1983a). Mapping the responses of MT cells with stimuli of different motion directions inside the RF typically reveals a Gaussian-shaped tuning curve. The peak of the curve, representing the strongest response, is centered on the neurons preferred direction. In contrast, motion directions opposite to the preferred direction usually produce weaker responses. This direction is commonly referred to as ‘anti-preferred’- or ‘null’-direction. A measure for selectivity in the directional tuning is the bandwidth, which is defined as the width of the tuning curve at half of the difference between the response to the preferred and antipreferred directions (Maunsell and Van Essen, 1983a). On average, response increments for motion in the neuron’s preferred direction are about four times the magnitude of response decrements for motion in its null direction (Snowden et al., 1991).
Most MT cells are also tuned to motion speed (Maunsell and Van Essen, 1983a). They are typically bandpass-tuned with a preference for intermediate speeds, while slower or faster speeds lead to response decreases. In addition, MT is sensitive to other aspects of visual information, such as stimulus orientation (Albright et al., 1984), binocular disparity (Maunsell and Van Essen, 1983b) and the direction of smooth pursuit eye movements (Komatsu and Wurtz, 1988).
The tuning properties of MT neurons suggest that this area plays a key role in the perception of visual motion signals. Indeed, several studies have demonstrated a close link between MT neuronal activity (or its absence) and perceptual experience.
For example, macaque monkeys with lesions in MT show impaired direction discrimination performance, while their contrast perception remains unaffected (Newsome and Pare, 1988; Pasternak and Merigan, 1994). Furthermore, the monkeys’
decision during direction discrimination can be biased by electrical stimulation of an MT direction column (Salzman et al., 1992), and there is a high degree of correlation between the neural threshold and the behavioral threshold for direction discrimination (Britten et al., 1992). This close relationship between neuronal and psychophysical threshold persists even when both vary across a recording session (Zohary et al., 1994).