Vera Katharina Veith1, Stefan Treue1,2
1Cognitive Neuroscience Laboratory, German Primate Center, Goettingen, Germany
2Bernstein Center for Computational Neuroscience, Goettingen, Germany
Abstract
Attending to one location in the visual field, guided by a pre-cue, leads to enhanced processing of the stimuli presented at that location, resulting in better performance and reduced reaction time (Treue & Maunsell, 1996; Posner, 1980). When the information about the subsequent target location is diminished or not provided at all, a behavioral cost is observed, in terms of a decrement in perceptual performance (Posner, 1980). Several psychophysical studies are not able to distinguish, whether the measured improvement in behavioral performance is ascribed to a reduction in stimulus uncertainty or to an actual improvement in information processing (Pashler, 1994; Pashler and Johnston, 1998). In this study, with the use of a novel task design, we are able to measure the pure influence of varying magnitudes of spatial attention on human
subjects’ performance in discriminating visual motion directions. An endogenous cue was used to gradually guide spatial attention to two stimuli, either placed across the two hemifields or within the left hemifield. Therefore we were additionally able to measure hemifield dependence on spatial attention capacities.
In the classical stimulus arrangement (across hemifield), observers were as good in simultaneously monitoring two spatial locations, than monitoring just one location. Even if deployed spatial attention was further diminished (25%
attentional deployment), subjects’ performance was not significantly influenced.
In contrast, when spatial attention was distributed within one hemifield, behavioral performance was deteriorated. This suggests that the decrement in performance is due to common processing resources, that the two stimuli are competing for when they are placed within one hemifield (Sereno and Kosslyn, 1991). These attentional capacity limitations seem to underlie anatomical constraints (Alvarez and Cavanaugh, 2005).
40 _______________________________________________________
Introduction
Spatial visual attention is a mechanism that upon deployment improves perceptual performance. Vision at an attended location is faster (e.g. Posner, 1980), more accurate, of higher spatial resolution (Yeshurun and Carrasco, 1998) and enhances sensitivity for fine changes (Carrasco et al, 2004). Not attended stimuli appear to have lower contrast or (if they are objects embedded in complex scenes) might not be perceived at all (Carrasco, 2006, Reynolds &
Chelazzi 2004).
Physiologically, attention has been reported to enhance neuronal responses in visual cortex to stimuli, when they are attended (Treue & Maunsell, 1996). Such a gain increase can increase neuronal performance and therefore offers a possible mechanism of attentional perceptual enhancement (Cook & Maunsell, 2002). Nevertheless, while the neuronal response enhancement to visual motion stimuli has been well-investigated and documented in area MT - an extrastriate cortical area thought to underlie the perception of linear motion - it remains controversial if and how much attention enhances the perception of basic visual features, such as visual motion (Liu, Fuller & Carrasco, 2006; Braun
& Julesz, 1998).
Additionally, it is still under debate how spatial attention is deployed in the visual field. In contrast to early studies showing the existence of only one spotlight of attention eliciting enhanced performance at that location (Posner et al., 1980, Eriksen & St. James, 1986), an increasing amount of recent studies argue in favor for the existence of (at least) two non-continuous regions in space that can be attended simultaneously (Awh & Pashler, 2000; Mueller et al., 2003). Moreover, it was reported that information of intervening regions could be ignored (Shaw & Shaw, 1977; Shaw, 1978; Awh & Pashler, 2000). Neuronal evidence supports this notion, showing that two spatially separated stimuli elicit
higher fMRI signal at the two corresponding retinotopic regions (McMains &
Somers, 2004, Morawetz et al., 2007).
As proposed by the hypothesis of the bilateral advantage of attentional distribution, attention can be deployed to two locations across the two hemifields without any cost being involved, whereas splitting attention within one hemifield uncover attentional resource limitations (Reardon et al., 2009;
Sereno and Kosslyn, 1991). The inter-hemispheric advantage is ascribed to the anatomical and functional independence of the two cerebral hemispheres, also showing full functionality, when the inter-hemispheric communication is disrupted by a surgical transection of the corpus callosum (Liederman, 1998, Luck et al., 1989). This means, when the two stimuli are presented in an across-hemifield arrangement, it is suggested that each stimulus is processed independently, whereas when both stimuli are placed within one hemifield, they seem to access the same source. However, visual field representations of extra striate cortical areas for the lower and upper part of the visual field are also often found to be anatomically segregated (Van Essen, 1985).
Here we investigate the effects of spatial attention on performance in a direction-discrimination task, using two moving dot patterns placed either across both hemifields or within the left hemifield. The overall aim of our study is to investigate the effects of various levels of spatial attention onto the two stimuli in dependence of their spatial configuration. Therefore, we aim to extent the bilateral attentional advantage, as previously shown for higher-level tasks (Alvarez and Cavanagh, 2005,) as well as for elementary visual tasks (Reardon et al., 2009), to the linear motion discrimination task.
In many cases, divided attention tasks suffer from larger stimulus uncertainty rather than the corresponding undivided attention (Posner et al., 1980; Pashler, 1994). We controlled for such effects, by using a four alternative forced choice (4AFC) task, allowing us to determine those trials in which performance suffered from errors due to stimulus uncertainty. We manipulated the validity
42 _______________________________________________________
of the cue in our paradigm across three values. This allowed us a fine-grained analysis of the effect of graded attention (100-25%) onto our stimuli. The effect of attention on performance is investigated by comparing subjects’ performance when various proportions of attention were allocated to a given stimulus.
With our chosen stimuli, we aim to target area MT, an extrastriate cortical area thought to underlie the perception of linear motion. As area MT is involved in early stages of cortical processing (Vanni et al., 2004), we were targeting bilateral vs. unilateral differences.
We demonstrate, that subjects can attend to two spatial locations across the two hemifields simultaneously without showing any cost in linear movement discrimination when spatial attention is gradually diminished. When stimuli were shown unilateral, splitting attention to two stimuli incorporates a cost and subjects showed a reduction in performance. Based on our results, we can support the notion of two independent attention systems, one being active on each hemifield, as we found bilateral attention advantage for linear motion direction discrimination tasks.