Funded Grants

Multisensory Neglect in Perception and Spatial Working Memory

Unilateral neglect is a disabling deficit of attention and spatial representation following unilateral brain injury, typically caused by strokes centred on the right inferior parietal lobe. The patients behave as if the contralesional side of space no longer exists, even though they are not strictly blind, deaf or insensitive to that side. Recent progress in understanding this disorder has been achieved by relating it both to functional models of normal human attention, and also (via computational modelling) to single cell data from parietal neurons in monkeys, for visual aspects of neglect (Driver & Mattingley, 1998; Pouget & Sejknowksi, 1997). However, there been much less work on how neglect affects non visual modalities, even though recent advances in crossmodal attention research with normals (see Driver & Spence, 1998a), and in studies of the multimodal representation of space at single cell levels (Andersen et al., 1997; Graziano & Gross, 1998), make the issue of crossmodal neglect particularly timely. Moreover, some of the multimodal single cell data, on spatial remapping of 'delay activity' across postural changes during spatial working memory, suggest a specific new account for how neglect should produce spatially-specific deficits in working memory. This can explain several otherwise puzzling aspects of the disorder.

We propose a series of experiments which exploit the recent developments in the psychological and single cell literature on crossmodal spatial attention, to assess their implications for neglect across multiple modalities. We will conduct systematic studies of whether certain lesions invariably produce neglect across vision, hearing, and touch, or whether neglect can be dissociated between these modalities. These studies adapt the methods we have recently developed and corroborated for the study of crossmodal spatial attention in normals, applying them to neurological patients for the first time. With further variations on these methods, we shall also test a counterintuitive hypothesis suggested by recent single--cell data and by computational arguments (see bundled Pouget proposal). On this hypothesis, the location of auditory and tactile stimuli gets remapped retinotopically in the brain, leading to the prediction that eye position should affect the severity of auditory and tactile neglect for a fixed unseen stimulus.

A further series of studies takes a different but complementary approach to crossmodal issues in neglect. These experiments examine whether the audiovisual integration processes that produce 'ventriloquism' (i.e. the illusion that a sound is located at the position of a temporally synchronous but spatially discrepant visual event) can influence auditory neglect. Specifically, we test whether pairing contralesional sounds with uninformative visual events further towards the ipsilesional side can reduce auditory neglect. In addition to their potential implications for rehabilitation, these studies address the key theoretical question of whether integration between the senses arises before the level at which neglect arises.

The final set of studies examines a new hypothesis for how neglect should be manifest within spatial working memory. Many physiological studies have documented neurons (in prefrontal, premotor and parietal areas) which show 'delay activity', representing a location that must be remembered across a brief retention interval. Recent single cell studies demonstrate that the peak of such activity can shift across internal maps (i.e. different units show delay activity sequentially) if part of the animal (e.g. the eyes, head or body) are turned during the delay period. A 'moving hill' of activity is seen, updating the current location of the remembered target relative to the animal as posture is changed. If, as widely believed, neglect is due to an impairment towards one side of such internal maps, then the location of remembered targets should be forgotten if the patient's posture is altered during the delay period, so as to move the 'hill' of delay activity into the damaged region of the map. This can explain several paradoxical features of the neglect syndrome, particularly the patients' tendency to neglect contralesional information which they had acknowledged only seconds previously. However, the prediction has never been tested before. Indeed, the neglect literature has not previously been integrated with the working memory literature, even though many of the patients' difficulties arguably lie in the formation of the more durable representations associated with working memory, rather than in transient sensory coding.