Is coherent motion an appropriate test for magnocellular sensitivity?

Abstract

The suggestion that coherent motion may serve as a test of magnocellular sensitivity is problematic. However, the nature of the problems depends on how the “magnocellular system” is defined. If this term is limited to subcortical entities, the problems are that subcortical neurons are not directionally selective, and that their receptive fields are too small to account for the spatial summation of coherent motion. If “magnocellular system” is defined to include cortical entities, such as area MT, one is faced with the fact that this definition itself is problematic as well as the problem that area MT is known to receive parvocellular and koniocellular inputs.

Introduction

The early part of the primate visual pathway contains three parallel streams: the magnocellular, the parvocellular, and the koniocellular systems (see Hendry and Reid, 2000, Merigan and Maunsell, 1993, Shapley and Perry, 1986, for reviews). The separation between these systems is most pronounced in the Lateral Geniculate Nucleus (LGN), where the magno- and parvocellular neurons form anatomically separate layers, with the koniocellular neurons occupying the interlaminar spaces (often referred to as koniocellular layers or K-layers; Hendry & Reid, 2000). From the LGN the magnocellular and parvocellular neurons send axons to two separate sublayers (layers 4C-α and 4C-β, respectively) of the primary visual cortex (V1). Although initial reports emphasized the separateness of magno- and parvocellular inputs in the cortex (Livingstone and Hubel, 1987, Livingstone and Hubel, 1988, Maunsell et al., 1990), it has become increasingly clear that it is difficult to maintain a clear magno-/parvocellular distinction beyond the input layers to the primary visual cortex, owing to the mixing of inputs inside the cortex (Lachica et al., 1992, Levitt et al., 1994, Martin, 1992, Merigan and Maunsell, 1993, Nealey and Maunsell, 1994, Sawatari and Callaway, 1996, Sincich and Horton, 2002, Vidyasagar et al., 2002; see also DeYoe and Van Essen, 1988, Dobkins and Albright, 2003, Kiper et al., 1999). The path of the koniocellular input to VI has yet to be worked out in detail; however, there is no evidence to suggest that this input gives rise to a separate cortical stream. It has recently been reported that in addition to providing input to V1, koniocellular LGN neurons send direct input (bypassing V1) to cortical area MT (Sincich, Park, Wohlgemuth, & Horton, 2004) (a cortical area considered by some to represent part of, or an extension of, the magnocellular system). Nevertheless, even this input does not constitute a completely pure koniocellular stream, since approximately one third of the neurons contributing axons to this input appear to be non-koniocellular (Sincich et al., 2004).

Starting in the primary visual cortex two cortical streams may be discerned: the “dorsal” and “ventral” streams. The dorsal stream includes the Middle Temporal area (area MT; also known as V5) and the ventral stream includes area V4 (Merigan and Maunsell, 1993, Ungerleider and Mishkin, 1982). The dorsal and ventral streams are linked to functional differences. The dorsal stream is widely believed to be implicated in motion perception (Livingstone and Hubel, 1987, Zeki, 1978a), specifically in perception of coherent motion (Stein, 2003), whereas the ventral stream is linked to the perception of color and form (Zeki, 1978b). The dorsal and ventral streams are sometimes referred to as the “where” and the “what” streams (Mishkin, Ungerleider, & Macko, 1983). Lately, some investigators have sought to extended the concept of the “magnocellular system” to include the dorsal stream (Stein, 2001, Stein, 2003) or to extend the “magnocellular” and “parvocellular systems” to include, respectively, the dorsal and ventral cortical streams (Edwards et al., 2004). There are (at least) three problems associated with extending the concepts of “magnocellular” and “parvocellular” in this way: (1) the nature of the distinction between the various systems is obscured, because the dorsal stream, in addition to magnocellular input, receives inputs which can be traced back to the parvocellular (Maunsell et al., 1990, Merigan and Maunsell, 1993, Sincich and Horton, 2002, Yabuta et al., 2001) and koniocellular systems (Sincich et al., 2004), and the ventral stream, judging from recordings in V4 following LGN lesioning, receives about equally strong inputs of magno- and parvocellular origin (Ferrera, Nealey, & Maunsell, 1994; see also Martin, 1992). (2) Extending the terms “magnocellular” and “parvocellular” in this manner reduces their predictive power given that neurons in the subcortical systems and in the visual cortex have quite different response properties. Thus, predictions based on subcortical magnocellular neurons would tend to be quite different from predictions based on neurons in the dorsal cortical stream. Finally, (3) a functional unity between the subcortical systems and the cortical streams is suggested or implied. Functional unity is problematic given the differences in the neuronal response properties and the differences in the constraints placed on the subcortical and cortical systems. For these reasons, we favor a definition of the “magnocellular” and “parvocellular systems” that restricts these terms to the subcortical portion of the visual system.¹ (It should be noted that when we use terms such as “magnocellular inputs” or “parvocellular inputs” in connection with higher order cortical areas, we are referring to neuronal responses whose origins can be traced back to, respectively, the subcortical magno- and parvocellular systems. We do not mean to imply that the magno- and parvocellular systems themselves extend to cortical areas beyond V1).

The functional significance of the magno-, parvo-, and koniocellular systems is not well understood. In attempts to gain insight into this issue investigators have sought to devise psychophysical and electrophysiological tests that selectively assess the functioning of one of the systems. These attempts have focused to a large extent on tests of the magnocellular system. However, many of these tests do not seem well suited to the task (Skottun, 2001a, Skottun, 2001b, Skottun, 2004a, Skottun, 2004b, Skottun and Skoyles, 2004). A number of investigators have claimed, suggested, or implied that perception of coherent motion can be used to assess the sensitivity of the magnocellular system, and consequently, that difficulties perceiving coherent motion may be attributed to deficits in the magnocellular system (Cornelissen et al., 1998a, Cornelissen et al., 1995, Cornelissen et al., 1998b, Eden et al., 1996, Edwards et al., 2004, Milne et al., 2002, Pammer and Wheatley, 2001, Schulte-Korne et al., 2004, Slaghuis and Ryan, 1999, Solan et al., 2003, Stein, 2001, Stein, 2003, Stein and Walsh, 1997, Talcott et al., 2000a, Talcott et al., 1998, Talcott et al., 2000b, Witton et al., 1998).

Perception of coherent motion is typically obtained by presenting a field of small moving random dots in which some portion of the dots move in the same direction, i.e., move coherently, while the rest move randomly. The percentage of coherently moving dots is then varied. The percentage of coherently moving dots required for the subject to perceive a coherently moving field of dots is taken as the measure of perceptual performance (Britten et al., 1992, Newsome and Pare, 1988). As will become clear below, the claim, or suggestion, that this performance should reflect magnocellular activity is problematic for several reasons.