Contrast induced changes in response latency depend on stimulus specificity

Abstract

Neurones in visual cortex show increasing response latency with decreasing stimulus contrast. Neurophysiological recordings from neurones in inferior temporal cortex (IT) and the superior temporal sulcus (STS), show that the increment in response latency with decreasing stimulus contrast is considerably greater in higher visual areas than that seen in primary visual cortex. This suggests that the majority of the latency change is not retinal or V1 in origin, instead each cortical processing area adds latency at low contrast. I show that, as in earlier visual areas, response latency is more strongly dependent on stimulus contrast than stimulus identity. There is large variation in the extent to which response latency increases with decreasing stimulus contrast. I show that this between cell variability is, at least in part, related to the stimulus specificity of the neurones: the increase in response latency as stimulus contrast decreases is greater for neurones that respond to few stimuli compared to neurones that respond to many stimuli.

Introduction

Neuronal response latency – the time when the stimulus elicited neuronal signal can first be detected – is an example of a neuronal code involving precisely timed spikes (see Oram et al., 2002 for review). Given the dissociation of response magnitude and response latency (Albrecht, 1995, Bair et al., 2002, Carandini and Heeger, 1994, Gawne et al., 1996, Reich et al., 2001a, Reich et al., 2001b) it is not surprising that response latency of individual neurons in primary visual cortex convey information unavailable from the spike count (Gawne et al., 1996, Reich et al., 2001a, Reich et al., 2001b). Indeed, it is has been speculated that changes in response latency are a potential source of the temporal code revealed by principal component and information theoretic analysis of spike waveforms (Optican and Richmond, 1987, Tovee et al., 1993). Thus, understanding factors that influence neuronal response latency may be relevant to studies examining the role of temporal variation in firing rate in visual processing (Eskandar et al., 1992, Heller et al., 1995, McClurkin et al., 1991, Optican and Richmond, 1987, Richmond and Optican, 1990) as well as shed light on the temporal precision of neuronal codes.

The latency of visually responsive neurones in the visual system increases with decreasing stimulus contrast in the retina (Shapley and Victor, 1978), LGN (Lee et al., 1981b), primary visual cortex (Albrecht, 1995, Carandini et al., 1997, Carandini et al., 2002, Carandini and Heeger, 1994, Gawne et al., 1996, Movshon et al., 1978, Parker et al., 1982, Reich et al., 2001a, Reich et al., 2001b, Wiener et al., 1998), area MT (Raiguel et al., 1999) and the anterior superior temporal sulcus (Oram et al., 2002, van Rossum et al., 2008). The increment in response latency with decreasing stimulus contrast is considerably greater in higher visual areas such as the anterior superior temporal sulcus (STSa) than that seen in primary visual cortex Fig. 1 and (Oram et al., 2002, van Rossum et al., 2008). Indeed, the average response latency in area STSa increases by 33 ± 3 ms for each halving of stimulus contrast compared to 8 ± 0.8 ms in V1 (van Rossum et al., 2008). The increasing dependency of neuronal response latency on stimulus contrast indicates that latency change is not retinal or V1 in origin, instead suggesting that each cortical processing area adds latency at low contrast (van Rossum et al., 2008).

Stimulus properties other than stimulus contrast influence response latency. For example, changes in spatial frequency influence both response magnitude and response latency of many V1 neurones (Bredfeldt and Ringach, 2002, Mazer et al., 2002). Position of moving gratings relative to the receptive field also influence response latency (Lee et al., 1981a), as does luminance of the stimulus (Maunsell and Gibson, 1992). On the other hand, changes in response magnitude do not necessarily influence latency in V1 (Albrecht et al., 2002, Gawne et al., 1996, Geisler and Albrecht, 1995, Opara and Worgotter, 1996, Reich et al., 2001a, Reich et al., 2001b, Tolhurst and Heeger, 1997, Worgotter et al., 1996). Similarly, response latency of individual neurons in STSa show little dependency on response magnitude (Oram et al., 1993, Oram and Perrett, 1996, Oram and Perrett, 1992) whereas changes in stimulus contrast cause large changes (>200 ms) in response latency (Oram et al., 2002, van Rossum et al., 2008, York et al., 2007).

In this article, I present data indicating that processing complexity may also influences neuronal response latency. Specifically, I show that the response latency of neurones that respond to a small number of stimuli is more sensitive to changes in stimulus contrast than neurones that respond in a less discriminative or selective fashion. The findings are discussed in terms of current models of visual processing.