Attentional and sensory effects of lowered levels of intrinsic alertness

Abstract

Low levels of intrinsic alertness are associated with lateralized performance in visual tasks, similar to neglect of the left (ipsilesional) visual hemi-field. However, it is unclear whether reduced alertness produces a specific lateralization of spatial-attentional processes in terms of the prioritization of right- over left-side stimuli, or whether it affects more basic functions of visuo-sensory coding, and/or higher function of the top-down control of selection, of stimuli on the left side. To decide between these alternatives, the present study examined the effects of lowered alertness, induced by a 50-min vigilance task, in a partial-report paradigm of briefly presented letter displays. With only one (unilateral) stimulus in display, no specific hemi-field effects were found under low-alertness conditions, indicating that reduced alertness impairs neither sensory effectiveness nor the top-town control of selection. However, with dual, bilateral stimuli, report accuracy was specifically affected for left-side targets (in subjects who showed comparable performance for both sides under normal-alertness conditions). This pattern can be interpreted in terms of a specific bias in spatial-attentional weighting, where prioritization of stimuli on the right leads to (mild) extinction of targets on the left. Moreover, participants who had a lower general level of alertness also showed a more pronounced re-distribution of weights, evidenced by a more severe imbalance in report accuracy, in a low compared to a normal state of alertness. This suggests that a low general level of intrinsic alertness engenders a specific vulnerability to neglect-like performance with a (mild) left-side extinction.

Introduction

Intrinsic (non-phasic) alertness refers to the internal or endogenous control of the level of arousal – in terms of the general readiness to respond sensory stimulation (Posner and Petersen, 1990, Sturm et al., 1999, Sturm and Willmes, 2001). Neuroimaging studies have revealed a widely distributed, predominantly right-hemispheric intrinsic-alertness network involving the frontal lobe, the inferior and superior parietal lobe, as well as thalamic and brain stem regions (Coull et al., 1998, Kinomura et al., 1996, Robertson et al., 1998, Sturm et al., 1999, Sturm et al., 2006, Sturm and Willmes, 2001). One of these areas, the right parietal cortex, plays an important role for attention (Kanwisher & Wojciulik, 2000) and spatial-attention functions in particular (e.g., Corbetta et al., 2000, Corbetta and Shulman, 2002). Consequently, space-based attention and intrinsic levels of alertness should be closely associated.

Behavioural evidence for such an association comes from studies of patients who suffer from a spatial-attentional asymmetry combined with a reduced level of (intrinsic) alertness, the most prominent example being patients with visual hemi-neglect who show a pathological ipsilesional spatial-attentional bias almost exclusively following right parietal lesions (Bisiach and Vallar, 1988, Heilman et al., 2003, Karnath et al., 2003, Vallar et al., 1994). A similar, though less pronounced pattern of spatial and non-spatial attentional deficits can be found in patients with attention deficit hyperactivity disorder (ADHD; Antrop et al., 2000, Carter et al., 1995, Dobler et al., 2005, George et al., 2005, Nigg et al., 1997, Sheppard et al., 1999, Tucha et al., 2006), which also seems to be related to a predominantly right-hemispheric (fronto-parietal) dysfunction. In both patient groups, the rightward deviation of spatial attention can be observed best in those individuals whose intrinsic-alertness state is especially low (Bartolomeo and Chokron, 2002, George et al., 2005, Heilman et al., 2003, von Cramon and Kerkhoff, 1993). Conversely, elevating the patients’ intrinsic-alertness state by, say, a warning stimulus (Robertson et al., 1998), stimulant medication (Nigg et al., 1997, Sheppard et al., 1999, Tucha et al., 2006), or alertness training (Robertson et al., 1995, Thimm et al., 2006) reduces their rightward bias at least temporarily. This effect is associated with increased activity within frontal and parietal brain regions of the right hemisphere (Thimm et al., 2006; see also Robertson et al., 1995).

The close connection between spatial attention and intrinsic alertness is not a ‘pathological characteristic’ of a damaged brain system. Similar relations are also present in healthy subjects. Under normal alertness conditions, healthy subjects tend to exhibit a slight leftward spatial bias referred to as ‘pseudo-neglect’ (Bowers & Heilman, 1980), which can be revealed, for example, in a line bisection task (for a review, see Jewell & McCourt, 2000) or the greyscales task (Mattingley, Bradshaw, Nettleton, & Bradshaw, 1994). For the latter task, Bellgrove, Dockree, Aimola, and Robertson (2004) reported the leftward pseudo-neglect bias to be significantly reduced in less alert healthy subjects, that is, subjects who performed poorly (compared to good performers) on a non-spatial alertness task. Several studies in young normal subjects have directly manipulated the intrinsic-alertness state by sleep deprivation, which is associated with decreased levels of global cerebral glucose metabolism especially in the alertness-related fronto-parieto-thalamic network (Kinomura et al., 1996, Thomas et al., 2000). For example, in a Posner-like exogenous cueing task, Fimm, Willmes, and Spijkers (2006) reported a disproportionate slowing of responses to stimuli presented on the left side of fixation for sleep-deprived subjects in a state of maximally lowered alertness, compared to responses of the same subjects in a normal alertness state. Similarly, Manly, Dobler, Dodds, and George (2005) found a significant rightward shift of spatial attention in sleep-deprived subjects using a landmark task (see also, Dufour, Touzalin, & Candas, 2007). These authors also observed a rightward shift of spatial attention over the course of the testing session (time-on-task effect). Interestingly, this spatial asymmetry could be significantly reduced by stimulant medication, that is, by increasing the (baseline) level of intrinsic alertness. Therefore, it has been proposed (Bellgrove et al., 2004, Fimm et al., 2006) that a reduction in the level of intrinsic alertness might suffice to induce a rightward shift of visuo-spatial attention even in subjects with a healthy attention system, who normally show a slight bias towards the left.

Despite this evidence, a closer understanding of the relationship between intrinsic alertness on the one hand and the spatial distribution of attention on the other is still lacking. Since healthy subjects with a relatively low baseline alertness show less leftward- (or even slightly rightward-) biased attentional performance (Bellgrove et al., 2004, Dufour et al., 2007, Manly et al., 2005), it remains unclear whether such subjects would be specifically prone to display unbalanced attentional performance. Especially under monotonous tasks conditions with only rare external phasic stimulation, which thus require an appropriate alertness state to be maintained internally (Posner and Boies, 1971, Posner and Petersen, 1990), subjects with low baseline levels of intrinsic alertness might show a more pronounced inattention to and extinction of left-sided stimuli compared to subjects with a higher baseline level.

Additionally, it remains an unsolved issue whether the altered performance in the left hemi-field under lowered intrinsic-alertness conditions (e.g., Bellgrove et al., 2004, Fimm et al., 2006, Manly et al., 2005) really reflects an attentional effect, rather than resulting from a change in sensory processing. In fact, in the majority of the studies reviewed above, no clear distinction was made between sensory and spatial-attentional factors. As (non-attentional) sensory and spatial-attentional aspects of performance seem to be confounded in these tasks, it is worth ascertaining whether changes in the intrinsic-alertness state become evident as a bias in sensory effectiveness or in spatial-attentional processes, or whether they occur in parallel with each other. One first piece of relevant evidence stems from Matthias et al. (in press), who demonstrated in a Posner-like cueing study that effects of cue-induced alertness changes can operate already at the sensory level. Matthias et al. (in press) proposed that (non-spatial) alerting cues influence sensory as well as spatial-attention processes, independently and in parallel. Given this, an assessment of alertness-related effects on (lateralized) sensory processing under conditions without spatial-attention demands is in order to ascertain whether and to what degree intrinsic alertness (independently) influences basic sensory effectiveness and spatial selection processes.

One way addressing these questions is based on the biased-competition view of visual attention (Desimone & Duncan, 1995), which assumes that sensory and spatial attentional processes can be disentangled by comparing performance with unilateral vs. bilateral stimulus displays: in unilateral displays, sensory effects are assumed to be independent of the distribution of spatial attention across the two hemi-fields. Any remaining side differences in the bilateral target condition should therefore be attributable to differences in the spatial distribution of attention. Thus, performance with bilateral target displays reflects the ‘pure’ spatial-attentional bias which is controlled for sensory factors, whereas performance in unilateral target conditions reflects the total processing rate for each hemi-field (rather than how spatial attention is distributed across the hemi-fields). Therefore, sensory lateralization should affect performance even in unilateral stimulus presentations, whereas attentional lateralization should become manifest only when left- and right-side stimuli compete for attentional resources (bilateral stimulus condition). In case of a rightward attentional bias, stimuli in the right hemi-field receive a higher attentional weight than competing stimuli in the left field, so that performance for the left hemi-field will decline when stimuli are presented bilaterally. Such biased attentional competition towards rightward stimuli would have no effect in a unilateral, but a strong effect in a bilateral target display – which would therefore constitute important evidence for the assumption that low intrinsic-alertness conditions exert a truly attentional effect on spatial processing. Moreover, it would correspond nicely to the behaviour found in patients displaying ‘extinction’. In extinction, a contralesionally presented stimulus is detected or identified relatively well when presented alone (i.e., without competing stimuli in the ipsilesional field), but the same stimulus is disregarded (‘extinguished’) or only poorly identified in the presence of simultaneously presented ipsilesional input (Bender, 1952). Thus, both the assumed rightward spatial bias under low intrinsic-alertness conditions and the extinction phenomenon are affecting the ability to detect and respond to left-sided stimuli, and, based on the biased-competition view of attention, can be assessed independently of basic sensory ‘deficits’.

Interestingly, the (posterior) parietal cortex is also heavily implicated in non-spatial, task-related aspects of attention, such as the efficiency of top-down control (Corbetta et al., 2000, Corbetta and Shulman, 2002, Desimone and Duncan, 1995, Friedman-Hill et al., 2003, Gehring and Knight, 2002, Hopfinger et al., 2001, Peers et al., 2005), which may also be affected by the state of intrinsic alertness (Coull et al., 1998). Thus, it is possible that the rightward lateralization found under low-alertness conditions is accompanied by an impaired (task-related) top-down control regarding the suppression of irrelevant information in the left visual hemi-field (see Duncan et al., 1999 for a similar argument). However, behaviourally, the evidence for an influence of the level of intrinsic alertness on task-related top-down control remains inconsistent and subject to speculation. For example, in neglect patients assumed to have a decreased state of intrinsic alertness, Snow and Mattingley (2006) showed intact goal-driven selection processes for stimuli in the contralateral hemi-field in a reaction time-based flanker task. By contrast, top-down control was not effective in limiting the processing of task-irrelevant features of ipsilesional stimuli. Snow and Mattingley therefore speculated that a lateralized bias in spatial attention leads to unselective prioritization of all features of stimuli appearing within the ipsilesional hemi-field, whether or not they are relevant to performing the task (note that this assumption of selection of all stimulus, whether task-relevant or not, is consistent with Duncan's (Duncan, Humphreys, & Ward, 1997) ‘integrated-competition hypothesis’). In contrast, other studies revealed, besides the expected ipsilesional spatial bias, a largely preserved top-down control of attention (e.g., Bublak et al., 2005, Duncan et al., 1999, Vuilleumier and Rafal, 2000).

Taken together, the present study was designed to assess whether an induced low-alertness state in normal subjects would lead to a specific bias in the spatial distribution of attentional weighting (rather than in sensory effectiveness and/or top-down control). Furthermore, we wanted to systematically assess whether the development of a rightward attentional bias under conditions of low intrinsic alertness depends on (i) the subject's baseline level of intrinsic alertness and (ii) his/her ability to maintain an appropriate intrinsic alertness state over time on task. In other words, would subjects with a low baseline level of intrinsic alertness (as indicated by slow RTs) and/or a poor ability to maintain an appropriate alertness state (as assessed by an increase in subjective sleepiness ratings and more directly via RT slowing over time-on-task) be prone to develop a rightward spatial lateralization and thus be at risk to develop extinction-like behaviour under (monotonous task) conditions inducing lowered levels of alertness?

To induce a state of lowered alertness, a 50-min visual vigilance task was used. Paus et al. (1997) showed that performing a vigilance task over such a long period results in a linear decrease of brain activity within the right-hemispheric fronto-parietal network responsible for maintaining an alert state, accompanied by an similarly linear increase of response latency.

In order to assess spatial-attentional alterations, we used a partial-report design with briefly presented letter displays. In this task, which is based on the biased-competition view of attention (see above) and Bundesen's ‘Theory of Visual Attention’ (TVA; Bundesen, 1990, Bundesen, 1998), subjects had to identify as many briefly presented ‘target’ objects (letters) as possible, where targets were defined as stimuli belonging to a pre-specified (colour) category; that is, subjects had to report stimuli that matched a particular selection criterion (targets: red letters), while ignoring stimuli that did not (distractors: green letters). Targets and distractors were presented in either the same (unilateral condition) or in opposite (bilateral condition) hemi-fields. From the observed probabilities of target identification, TVA permits separate estimates of attentional weights to be derived for targets ($w_T$) and distractors ($w_D$), for each visual hemi-field ($w_{left}$ and $w_{right}$). Intuitively, objects with high attentional weights are processed relatively well, but interfere strongly with other objects, whereas objects with low attentional weights are processed poorly, and interfere weakly with others.

Via mathematical, TVA-based modelling of partial-report performance, quantitative and independent estimates of ‘bottom-up’ sensory effectiveness (in the left and the right hemi-field), as well as of two types of weighting were obtained: the spatial distribution of attentional weighting (spatial weighting of objects in different parts of the visual field) and the task-related weighting, that is the effectiveness of top-down controlled selection. The estimates of the two weighting parameters were based on differences between accuracy for a single target and for the same target accompanied by other objects either targets or distractors on the same or the opposite side. A detailed formal description and the equations of TVA as well as information on the model fitting procedure, which is based on a maximum-likelihood procedure, can be found in Kyllingsbæk (2006). The partial-report paradigm has been shown to be sensitive for even subtle changes in spatial attentional weighting that are not detectable by conventional procedures (Finke et al., 2006, Habekost and Bundesen, 2003). Furthermore, it was shown that even in healthy subjects those with a spatial attentional bias were slower on a visual scanning task than those with an optimally balanced distribution of weights across both visual hemi-fields (Finke et al., 2005). Thus, it is a suitable instrument to differentiate between slight attentional laterality differences that can be expected between healthy subjects in different arousal states. The possibility to address spatial attentional weighting effects specifically and to disentangle them from bottom-up sensory as well as from top-down controlled selection effects furthermore allows to draw specific conclusions on whether lateralized performance is indeed induced by a rightward lateralization of spatial attentional weights.