Professional Documents
Culture Documents
Neuropsychologia
journal homepage: www.elsevier.com/locate/neuropsychologia
a r t i c l e i n f o a b s t r a c t
Article history: In visual hemi-neglect, non-spatial decits such as reduced intrinsic alertness can signicantly modulate
Received 10 January 2012 the degree of left visual eld inattention. However, to date, the precise mechanisms mediating this effect
Received in revised form 6 February 2012 are hardly understood. In the present study, we assessed the inuence of increased alertness on both gen-
Accepted 15 February 2012
eral attentional capacity (perceptual processing speed) and spatial attentional selection processes (spatial
Available online 23 February 2012
distribution of attentional weighting). For this purpose, a whole-report paradigm based on Bundesens
theory of visual attention (TVA) was combined with a non-spatial, visual alerting cue. Three different
Keywords:
cue-target stimulus onset asynchronies (SOAs; of 80, 200, and 650 ms), allowed us to observe the time
Attention
Psychophysics course of the alerting-cue effects. A group of six patients with visual hemi-neglect was examined and
Intrinsic alertness their performance compared with six healthy control subjects matched for age, gender, and education.
Phasic alertness In neglect patients, the alerting cue evoked a phasic increase of perceptual processing speed. However,
Thalamus this effect was mainly found in the ipsilateral, i.e. in the preserved hemield. Importantly, however,
patients displayed a fast-evolving and short-lasting, phasic modulation of spatial attentional weighting,
with a re-distribution of attentional weights from the pathological rightward bias to a normal, more bal-
anced distribution of visual attention. In control participants, the cueing effects on perceptual processing
speed and spatial weighting were generally less pronounced than in neglect patients. Replicating results
of a prior study, cueing induced a stable, slightly leftward, distribution of attentional weights, whilst in
the no-cue condition, a temporary rightward shift of attentional weights was found.
This pattern of effects suggests a close interaction between alertness and spatial-attentional weighting
in the syndrome of visual hemi-neglect. It supports the hypothesis that the manifestation of spatial neglect
involves at least in part intrinsic alertness decits. It also provides clues to a more detailed account of
the mechanisms responsible for alleviating neglect in patients following manipulations of the alertness
level, both in the short (cueing) and in the long term (alertness training).
2012 Elsevier Ltd. All rights reserved.
1. Introduction of awareness for those in the left hemi-eld is the most promi-
nent feature of neglect (Bartolomeo & Chokron, 2002; Bisiach
It is now well established that both spatially lateralized and & Vallar, 1988; Heilman, Watson, & Valenstein, 2003; Karnath,
non-lateralized decits of attention contribute to the clinical pic- 1988), patients also show non-spatial decits such as, for exam-
ture of left-sided visual hemi-neglect (Husain & Rorden, 2003; ple, a prolonged attentional blink (Husain, Shapiro, Martin, &
Samuelsson, Hjelmquist, Jensen, Ekholm, & Blomstrand, 1998; van Kennard, 1997), impaired working memory (Danckert & Ferber,
Kessel, van Nes, Brouwer, Geurts, & Fasotti, 2010). Although, at 2006; Wojciulik et al., 2001), and decient vigilance (Hjaltason,
the behavioural level, a spatially lateralized preference for stim- Tegnr, Tham, Levander, & Ericson, 1996; Robertson et al., 1997).
uli and objects on the right side of space combined with a lack In fact, these non-spatial symptoms may even be more important
for the severity and persistence of hemi-neglect than the spatial
bias itself (Husain & Rorden, 2003). They may result from reduced
alertness, a foundational form of attention or processing capacity,
Corresponding author. Tel.: +49 0 89 218072520. which more complex cognitive functions draw from (Raz & Buhle,
E-mail address: nke@psy.lmu.de (K. Finke). 2006).
0028-3932/$ see front matter 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.neuropsychologia.2012.02.008
K. Finke et al. / Neuropsychologia 50 (2012) 11781189 1179
Lowered alertness, i.e. an impeded ability to achieve and main- interaction between alertness, non-spatial and spatial attention
tain a state of high sensitivity to an impending stimulus (Posner, in healthy subjects. To that end, whole-report performance was
2008; Raz & Buhle, 2006), is a common observation in neglect compared between two conditions: a no-cue condition, assumed
patients (Heilman et al., 1978). Whilst tonic alertness refers to to reect possible changes in intrinsic alertness over the course of
the intrinsic and more long-term control of the arousal level inde- a trial (Coull & Nobre, 1998; Posner, 1978; Sturm et al., 1999), and
pendent from external warning cues, phasic alertness denotes the an alerting-cue condition, assumed to reect modulations of phasic
brief adaptive increase of the arousal level pending on an upcom- alertness. In this way, it was possible to examine whether different
ing warning stimulus (Posner, 2008; Raz & Buhle, 2006; Sturm levels of alertness would affect non-spatial (i.e. processing speed)
& Willmes, 2001). Functional brain imaging studies have shown and spatial attentional components differentially or, alternatively,
a substantial overlap between the cerebral networks underlying in a global manner. In addition, by introducing different cue-target
tonic and phasic alertness on the one hand, and those engaged in SOAs, it was also possible to map the time course of these effects,
governing spatial attention on the other, with the right inferior pari- so as to ascertain whether alertness-related effects on non-spatial
etal cortex, the most frequent lesion site in neglect patients, being and spatial attention develop in a parallel or a sequential manner.
a key node in both (Coull, Frackowiak, & Frith, 1998; Kinomura, Matthias et al. (2010) found that an increase in the level of alert-
Larsson, Gulyas, & Roland, 1996; Robertson, Mattingley, Rorden, & ness irrespective of whether this was phasically or intrinsically
Driver, 1998; Sturm & Willmes, 2001; Sturm et al., 1999, 2006). induced rst enhanced general processing speed and was then
In congruence with these overlapping brain networks of attention, followed by a leftward shift of spatial attention. They interpreted
less alert neglect patients have been shown to have greater leftward these results as indicating that intrinsic and phasic alertness effects
inattention (Bartolomeo & Chokron, 2002; Funk, Finke, Mller, Utz, involve the same processing route, on which spatial and non-spatial
& Kerkhoff, 2010; Heilman et al., 2003; von Cramon & Kerkhoff, mechanisms of attention are mediated by independent processing
1993). The alertness and spatial attention networks represent dis- systems which are activated, as a result of enhanced alertness, in
sociable brain systems (Husain & Rorden, 2003), however, and the temporal succession.
current knowledge about the nature and dynamics of their interac- Based on these preceding results in healthy young subjects, we
tion is only sparse. The present study is aimed at investigating this applied the same procedure in neglect patients. Six patients with
question. left visual hemi-spatial neglect following right temporo-parietal
Robertson et al. (1998) were the rst to directly examine in lesions and six healthy normal control subjects were assessed with
neglect patients, on a trial-by-trial basis, the causal relationship a whole-report task, in which they had to name stimuli presented
between alertness changes induced by sudden onset tones and either unilaterally, on either side of the visual eld, or bilaterally, in
modications of the rightward spatial bias. The authors found that the left and the right hemi-eld. As in the prior study on normal sub-
the pathological asymmetry of spatial attentional selection indeed jects (Matthias et al., 2010), the whole-report task was presented in
was alleviated by such a phasic enhancement of alertness. The alert- a no-cue and an alerting-cue condition. SOA variations were used
ing cues temporarily reduced the tendency of neglect patients to to map the time course of the effects of the alertness level on an
report the rightmost of two bars as coming rst in a temporal order index of processing speed (i.e. sensory effectiveness) and spatial
judgement task. This result demonstrates, that even the severely attentional weighting. We expected to nd similar results in normal
damaged attentional system of neglect patients bears the potential control participants as in our preceding study and wanted to anal-
to rapidly adapt in response to a warning cue so that their chronic yse, which of the processes and mechanisms prevalent in normal
inattention to the left hemi-eld can be overcome, resulting in a subjects would be altered or impaired in neglect patients.
more balanced spatial lateralization. Before presenting our method and the results, however, we pro-
However, the underlying mechanisms remain unclear. A closer vide, for those readers unfamiliar with TVA, a brief description of
understanding of the nature and time course of the effects of alert- its main ideas and concepts.
ness changes on non-spatial and spatial attention could, though,
provide important clues as to the cognitive architecture of attention 1.1. Theory of visual attention (TVA)
networks. For example, the exact relationship between phasic and
intrinsic alertness is still under debate (Posner, 2008; Raz & Buhle, In TVA (for a detailed mathematical description see Bundesen,
2006) and the question whether they are supported by the same 1990, 1998; Duncan et al., 1999; Kyllingsbaek, 2006), visual objects
or by different systems remains unresolved. Furthermore, closer and their features are assumed to be processed in parallel and
knowledge about alertness effects could also have implications for compete for selection. The race amongst objects and features
therapeutic intervention. Current treatment approaches suggest to is decided according to a speed criterion: features which enter
be based on the modulation of intrinsic alertness (Thimm, Fink, visual short term memory (VSTM) rst cause the complete encod-
Kust, Karbe, & Sturm, 2006) or both phasic and intrinsic aspects ing of these objects into VSTM. Biases can arise if some objects
(DeGutis & van Vleet, 2010), whilst clear evidence about the rele- receive higher attentional weights than others due to either auto-
vant mechanisms mediating improvement in neglect is still lacking. matic (bottom-up) or intentional (top-down) factors, conferring
To investigate this issue, we used an approach based on a for- a speed advantage in the race for selection. According to TVA, the
mal theory of visual attention (TVA, Bundesen, 1990, 1998) that probability with which an object is selected (when VSTM is not yet
allows simultaneous assessment of non-spatial and spatial aspects lled up to capacity) is thus determined by (i) the general process-
of attention within the same subjects and with the same task. In ing capacity available reected by an objects basic processing
neglect patients, such a method had been applied for the rst time rate (speed) and (ii) the attentional weight assigned to an object.
by Duncan et al. (1999). These authors used whole and partial report Independent quantitative estimates of these two atten-
of briey presented letter arrays to obtain parametric estimates of tional components general processing capacity and attentional
perceptual processing speed as a non-spatial component and lat- weight are derived from the patients whole-report performance.
eral bias as a spatial component of visual selective attention. They One aspect of weighting in TVA, which is especially relevant to
found, conrming the presence of non-spatial attentional decits the present study, concerns the lateral distribution of attentional
in neglect, processing speed to be signicantly reduced in patients weights within the visual eld. In TVA, the spatial distribution of
within both visual hemi-elds. attention across the visual eld is indicated by a spatial laterality
In a recent study, Matthias et al. (2010) used an analo- index (parameter w ), relating the attentional weights for objects
gous methodology to assess the nature and time course of the in the left and the right visual hemi-eld (wleft and wright ). These
1180 K. Finke et al. / Neuropsychologia 50 (2012) 11781189
Fig. 1. Lesion locations for neglect patients (PB, OB, KKL, ML, FP, and EW, from top to bottom) reconstructed for 8 transversal slices (left-hand side) and their positions in
sagittal orientation (right-hand side); numbers above each slice document the z-score in Talairach coordinates and lesion overlap (at the very bottom); the highest overlap
of patient groups lesions is shown in light red, the lowest overlap in dark red.
no emphasis on reporting speed. The experimenter entered the reported letter(s) outside this range, the exposure duration in the experimental phase was adjusted
using the computer keyboard and initiated the next trial after the participants had accordingly. The individual exposure durations of each participant are given
indicated that they were ready. The intertrial interval was 1000 ms. in Table 3.
The target exposure duration was individually determined for each participant
in a pretest part and then introduced into the experimental phase. The pretest con- 2.2.2. Alertness conditions
sisted of 72 single target trials (without alerting cues), presented for an exposure We compared an alerting-cue condition (assumed to involve a phasic alert-
duration of 171 ms. It was used to determine whether a participant was able to ness enhancement for a period of a few hundred milliseconds) with a no-cue
reach an accuracy of 6080% for single-target report. If the participant performed condition (purely involving intrinsic alertness). In the alerting-cue condition,
1182 K. Finke et al. / Neuropsychologia 50 (2012) 11781189
Fig. 2. Sequence of frames presented on a no-cue trial (top panel) and an alerting-cue trial (middle panel), together with the eight possible target displays: single targets,
dual targets in the same hemi-eld and dual targets presented in opposite hemi-elds (bottom panel; the T symbols denote target locations).
a white outline frame (5 5 ) ashed briey around the whole (potential) dis- location is likely to have discouraged patients from making eye movements. In any
play array. In the no-cue condition, the screen remained blank for the same length case, because the stimulus exposure durations were relatively short (200 ms), eye
of time. movements were unlikely to affect performance systematically. However, espe-
The alerting cue was non-informative as to the location of the upcoming target cially in neglect patients, suboptimal xation might occur already at the beginning
letters. Thus, although alerting the participants to the imminent appearance of the of a trial. Thus, to better control for eye- and head-movements during testing,
target array, this warning signal was designed to induce a spatially diffuse distri- we used a light-sensitive web-cam. With this device, eye and head movements
bution of attentional weighting across the (potential) stimulus display (i.e. it could could be observed on-line by the experimenter with minimal distraction of the
not be used to systematically orient spatial attention to the stimulus locations). participant. When eye or head movements occurred, participants were reminded
The non-informativeness of the alerting-cue with regard to the upcoming target to hold the xation and to try to avoid further movements.
induce a more general alerting/arousing effect, rather than supporting any specic 3. Results
temporal expectations about the onset of the stimulus array.
In what follows, we will rst describe the qualitative accuracy
2.2.4. Procedure pattern (correctly identied letters) obtained across the different
The PC-controlled experiment was conducted in a dimly lit room. Stimuli were groups and experimental conditions. Second, we will present the
presented on a 17 in. monitor (1024 768 pixel screen resolution, 70 Hz refresh
parameter estimates derived in TVA-based model ts to the data,
rate). Participants viewed the monitor from a distance of 50 cm, controlled with the
aid of a head- and chinrest. that is, sensory effectiveness and spatial attentional weighting.
Fig. 2 illustrates the sequence of frames presented on a no-cue (top panel) and Next, in order to compare sensory effectiveness and spatial atten-
an alerting-cue (bottom panel) trial. Following initiation of a trial sequence (by the tional weighting effects across groups, SOAs and cueing conditions,
experimenter), participants xated a white xation cross (0.3 0.3 ) which was mixed ANOVAS were carried out on the TVA parameters in ques-
presented for the entire trial duration in the centre of the screen, on a black back-
ground. Then, after 1100 ms, either the alerting cue appeared for 50 ms or the screen
tion.
remained blank for the same length of time. After the variable cue-target SOA, the
whole-report display was presented for the pre-set exposure duration. Following 3.1. Report accuracy: raw data indicators of sensory
registration of the participants report, the next trial started after an intertrial inter- enhancement and of changes in the spatial distribution of
val of 1000 ms. Cue/no-cue conditions as well as the three SOAs were presented in
attention
random order within the same block.
In the rst session (lasting around 30 min), the whole-report pre-test was
applied to determine the individual presentation times. In neglect patients, fur- To illustrate sensory effectiveness and attentional weighting
thermore, the BIT was performed. The second to the fourth session (lasting around across the two hemi-elds, Fig. 3 shows separately the mean pro-
45 min each) consisted of the experimental phase of the whole report. It com- portion of target letters correctly identied by patients and controls
prised of eight different target conditions (four single-target and four dual-target
in each hemi-eld and at each SOA for ve experimental conditions:
conditions) for each SOA (80, 200, 650 ms) and each of the two cueing conditions
(no-cue, alerting-cue). In total, the experiment comprised 846 trials per subject. single-target letter; target accompanied by a target in the ipsilat-
Standardized verbal instructions were given before each block of trials. In order to eral eld (column displays) and target accompanied by a target in
avoid or minimize a possible inuence of the alerting cue on the no-cue condition, the contralateral eld (row displays). In general, the performance
alerting-cue and no-cue trials were presented in different blocks. The order of the
levels of both groups are in accordance with the prediction from
blocks and sessions was counterbalanced across subjects to control for sequence
effects. Hence, for three patients and three control subjects, the sequence was AB, the TVA model: accuracy was highest in single-target conditions
BA, AB, and for the other three subjects of each group BA, AB, BA (with A = block and decreased in dual target conditions.
with no-cue trials, and B = block with alerting-cue trials). To control for time- Accuracy in single target conditions (rst two bars of each g-
on-day effects, the participants completed each of the four sessions at the same ure) can be taken as a raw data indicator of how well a stimulus is
time of day and within one week. A 5-min break was included in the middle of
basically processed, at a given exposure duration, under conditions
each session.
of no stimulus competition, that is, of pure sensory effectiveness, in
each hemi-eld. Due to individually adjusted exposure durations
2.2.5. Estimation of the TVA parameters
for each participant, between-group comparisons do not make
The whole-report task permitted quantitative estimates to be derived of visual
sensory effectiveness A and of attentional weight w, applying the algorithms sense with respect to the general level of accuracy. However, accu-
described in detail by Kyllingsbaek (2006) and used in several recent studies (e.g. racy changes across alerting conditions and hemi-elds can be
Bublak et al., 2005; Duncan et al., 1999; Finke et al., 2005; Habekost & Bundesen, taken as indicators of changes of sensory effectiveness and of spa-
2003). These parameters were estimated separately for the four positions used in
tial weighting. Fig. 3a shows that in neglect patients an alerting
the task (upper left, upper right, lower right, lower left): parameters Ai and wi ,
respectively, where i is the stimulus position.
cue seems to enhance accuracy for single targets at the 80-ms and
Enhancement of global processing capacity of a given participant by an alert- 200-ms SOAs. However, this improvement does mainly occur in the
ing cue would be reected in higher estimates of A across all stimulus positions. right hemi-eld; only at the longest, 650-ms SOA, an improvement
Indices of sensory effectiveness were calculated for each hemi-eld (by averaging is also manifest in the left eld. In control participants, an alertness
across the respective values for the upper and lower positions within each hemi-
cue does not seem to enhance single-target report in general. How-
eld): Aleft and Aright . This permitted us to examine whether alerting cues enhanced
sensory processing globally (across hemi-elds) in neglect patients, or whether, ever, also in this group, an enhancement is evident for left visual
possibly, sensory processing in the neglected contralesional hemi-eld would par- eld single targets at the longest SOA (Fig. 3b).
ticularly benet (and, thus, would specically impact the lateralized disturbances A comparison between accuracy losses in column displays on
in the syndrome).
the one hand and row displays on the other indicates possible
The parameter attentional weight is related to performance losses in multi-
element displays. Stimuli with higher weight are stronger competitors and, thus,
changes in spatial attentional weighting processes. As would be
disturb report accuracy for further stimuli more than stimuli with lower attentional expected in neglect patients with a reduced awareness of the
weight. The estimated attentional weights for upper and lower stimulus positions left side, in the no-cue condition, report accuracy was lowest for
were averaged in order to receive separate indices of attentional weighting for each row displays (i.e. in conditions with competition between the two
hemi-eld, wleft and wright . However, the relative, rather than the absolute, amount
hemi-elds), especially for the left side. However, this pattern was
of attentional weight attributed to each hemi-eld provides useful information on
spatial attentional selection. Thus, a laterality index the parameter spatial distri- modied in the alerting-cue conditions. Especially at the shortest
bution of attentional weighting (w ) was computed. This index (which is related SOA, report accuracy for the left side even exceeded that for the
to performance losses with (row) multi-element- compared to single-element dis- right side (Fig. 3a).
plays) indicates whether one hemi-eld is favoured in the spatial selection process.
If attentional weights are biased towards one hemi-eld, performance in bilateral
(compared to unilateral) presentation conditions will suffer more for targets pre-
3.2. TVA parameter estimates
sented in the hemi-eld with relatively low attentional weight, compared to targets
in the hemi-eld with high weight. Formally, this index is dened as the ratio of For each participant, cueing condition and SOA, sensory effec-
values across the two hemi-elds, w = wleft /(wleft + wright ). Hence, a value of w = .5 tiveness and attentional weights were quantied for each of the
indicates a balanced distribution of weights, values of w > .5 indicate a leftward,
four possible target locations. The empirically assessed mean accu-
and values of w < .5 a rightward spatial-attentional bias. If attentional weights are
biased towards one hemi-eld, performance in bilateral (compared to unilateral) racy scores for the different experimental conditions and the values
presentation conditions will suffer more for targets presented in the hemi-eld theoretically predicted (based on the best ts of the TVA model
with relatively low attentional weight, compared to targets in the hemi-eld with parameters) showed a close correspondence: the average corre-
high weight. Note, though, that under bilateral stimulus conditions, it is in prin- lation between the observed and the predicted scores across all
ciple also possible that a difference in performance between left- and right-side
targets is due to reduced sensory effectiveness in target discrimination on one
SOAs in the no-cue condition was .92 (neglect patients, SD = .04) and
side. This factor was controlled for by the hemi-eld-specic analyses of sensory .86 (control participants, SD = .07). In the alerting-cue condition, it
effectiveness A. was .88 (neglect patients, SD = .09) and .83 (control participants,
1184 K. Finke et al. / Neuropsychologia 50 (2012) 11781189
Accuracy (%)
Accuracy (%)
80 80
L L
60 60
R R
40 40
20 20
0 0
Accuracy (%)
80 80
60
L L
60
R R
40 40
20 20
0 0
ST DT column DT row ST DT column DT row
100
100
Accuracy (%)
Accuracy (%)
80 80
60
L L
60
R R
40 40
20 20
0 0
100 100
Accuracy (%)
Accuracy (%)
80 80
60
L 60
L
R R
40 40
20 20
0 0
ST DT column DT row ST DT column DT row
Accuracy (%)
80 80
60
L L
60
R R
40 40
20 20
0 0
ST DT column DT row ST DT column DT row
Accuracy (%)
80 80
60
L 60
L
R R
40 40
20 20
0 0
ST DT column DT row ST DT column DT row
Fig. 3. (a and b) Mean proportions of correctly identied target letters (% correct) presented in either the left or the right hemi-eld in the different experimental conditions,
separately for neglect patients (a) and for control participants (b). Each bar represents the mean in one condition: single target (ST), target accompanied by a target in the
same hemi-eld/dual targets in column displays (DT column), target accompanied by a target in the opposite hemi-eld/dual targets in row displays (dual target row). For
each group, separate gures are shown for the two cueing conditions (no-cue, alerting-cue) and the three SOAs (80, 200, 650 ms). Error bars give the standard errors.
K. Finke et al. / Neuropsychologia 50 (2012) 11781189 1185
Sensory effectiveness A
Sensory effectiveness A
5 5
4 4
3 no-cue 3 no-cue
2 cue 2 cue
1 1
0 0
80 200 650 80 200 650
SOA SOA
Sensory effectiveness A
5 5
4 4
3 no-cue 3 no-cue
2 cue 2 cue
1 1
0 0
80 200 650 80 200 650
SOA SOA
Fig. 4. Parameter A (sensory effectiveness) for the left (left half) and the right hemi-eld (right half), separately for neglect patients (upper half) and control participants
(lower half). The different lines depict sensory effectiveness for the two cueing conditions (no-cue, alerting-cue) as a function of SOA (80, 200, 650 ms). Error bars give the
standard errors.
SD = .04). Thus, in the no-cue condition, the model explained 84% separately for the two hemi-elds. As can be seen, the temporary
(neglect patients) and, respectively, 74% (control participants), and sensory enhancement in the right eld is reliably observed at the
in the alerting-cue condition it explained 77% (neglect patients) 80-ms and 200-ms SOAs, in each single case. Cue-induced sensory
and, respectively, 68% (control participants) of the variance. effects in the left, neglected, eld are generally more variable across
4. Discussion
The aim of the present study was to investigate the nature and
time course of the effect of a phasic alerting cue on spatial and
non-spatial aspects of attention, as compared to a no-cue condi-
tion, in patients with unilateral neglect for contralesional space. By
using a TVA-based whole-report paradigm our study was designed
to independently and separately examine the inuence of cue-
induced phasic alertness on processing capacity (i.e. on sensory
effectiveness as an indicator of processing speed) and on the spa-
tial distribution of attentional weights across the two hemi-elds
(i.e. on the laterality of selection). We assessed (1) whether cueing
would affect sensory and/or spatial attentional-components and, if
so, (2) in what time ranges these various effects would occur and
Fig. 7. Single case data of parameter w (spatial distribution of attentional weight- (3) whether they would differ from those found in normal control
ing) for each of the neglect patients in the no-cue condition (upper gure) and participants.
the alerting-cue condition (lower gure), as a function of SOA (80, 200, 650 ms).
As a rst main result of our study, control subjects showed
Error bars give the standard errors. Values of w > .50 = leftward attentional bias;
w < .50 = rightward attentional bias; w = .50 = no bias.
the greatest effects in the no-cue condition, with complemen-
tary, even opposing patterns for sensory effectiveness (left-eld
enhancement) and spatial weighting of attention (rightward shift),
especially at the medium SOA. This result is in close correspondence
in each patient and cueing condition. Following an alerting cue, with the data presented by Matthias et al. (2010) in their Experi-
the overall pattern described above at the group level was con- ment 2, which had applied the same paradigm as in the present
sistently present in each single case. Each patient showed a study, in particular using a randomised variation of SOA, in young
lateral redistribution of attentional weights towards the left side, healthy subjects. Matthias et al. (2010) interpreted this result as
and even a slight leftward bias in each single case (w > .5) reecting the effects of top-down mediated modulations of intrin-
at the shortest, 80-ms SOA which gradually disappears as SOA sic alertness and conrmed this assumption by showing that the
increases. pattern disappeared when SOAs were presented blockwise, so that
In summary, cueing had a marked impact on the spatial bias of adaptation of a mental set on a trial-by-trial basis was no longer
attention in neglect patients. In fact, at the shortest SOA, the right- necessary. The seemingly contradictory overlay between enhanced
ward bias disappeared and turned into a leftward spatial weighting intrinsic alertness and a, in relation to the other SOAs, more right-
of attention. This effect dropped linearly at the later SOAs, so that at ward spatial bias at the medium SOA, was interpreted as indicating
the longest SOA, a rightward bias was again obtained. In contrast, that enhanced alertness inuences two separate systems one sup-
more or less the same rightward bias prevailed across all SOAs in porting processing speed, the other governing spatial attention
the no-cue condition. that act independently from each other, but on a different time
scale. The enhancement of processing speed occurs rst, and is
followed by a leftward shift of spatial attention, observable at the
3.4.2. Control participants longest SOA.
The main effect of cue was not signicant [F(1, 5) = 1.61; p < .01], Following this assumption, the results of the control subjects in
but the ANOVA revealed a signicant main effect of SOA [F(2, the present study can be interpreted as further evidence that spatial
4) = 8.31; p < .05] and a tendency for an SOA cue interaction [F(2, and non-spatial mechanisms of attention are mediated by inde-
4) = 3.32; p < .08]. Mainly driven by the no-cue condition (see Fig. 6), pendent processing systems which exert their effects, as a result of
a slight but signicant shift of spatial attentional weights from the enhanced intrinsic alertness, in temporal succession. According to
left to the right side occurred from the shortest SOA of 80 ms to that the neuro-cognitive model proposed by Heilman et al. (2003) and
of 200 ms (t(5) = 3.60; p < .05) and another signicant shift back to Posner and Petersen (1990), modulations of intrinsic alertness are
the left side from the SOA of 200 ms to that of 650 ms (t(5) = 4.27; mediated by a right-hemispheric network in which frontal regions
p < .01). The spatial distribution of attention did not differ signif- exert top-down control, via thalamic nuclei, on posterior cortical
icantly between the shortest and the longest SOA (p > .75). In the areas related to perceptual processing, by activating noradrenergic
cue condition, there was a slight leftward shift of spatial attentional nuclei of the ponto-mesencephalic part of the brainstem (see also
1188 K. Finke et al. / Neuropsychologia 50 (2012) 11781189
Mesulam, 2000, for a similar idea). Consistent with right temporo- which is directly related to phasic alertness and responds to
parietal damage interfering with this system, and in congruence the alerting cue. A possible candidate for this would be the
with empirical evidence that regulation of intrinsic alertness is de- ascending reticular activating system, which receives direct sen-
cient in neglect patients (Bartolomeo & Chokron, 2002; Heilman sory input (Jones, 2003). These assumptions would predict that
et al., 2003), there was almost no sign of any modulation of the neglect patients with thalamic lesions do not show effects of
performance of neglect patients in the no-cue condition of our cue-induced enhancement of phasic alerting, and, therefore, no
study. increase of sensory effectiveness and no balanced spatial weighting
In contrast, and as the second main result of our study, should be observed in these patients after presenting an alertness
neglect patients showed the greatest effects in the cue condi- cue.
tion, especially at the shortest SOA. Interestingly, as for control Our data clearly show that enhanced phasic alerting brings
subjects though complementary concerning the hemi-eld the about a re-distribution of attentional weights in neglect patients.
pattern for sensory effectiveness (right-eld enhancement) and Whilst a rightward spatial bias prevailed in the no-cue condition
spatial bias (leftwards) was opposing. Especially because the sam- and also at the longest SOA in the cue condition, a more balanced, in
ple size was relatively small in our study, it is noteworthy that this fact even slightly leftward spatial bias was found 80 ms and 200 ms
result was highly consistent across patients at the single case level. after cue presentation. This effect was very consistent on the single
This pattern is compatible with the above presented interpretation case level and was found in each of the six patients assessed (see
of separate systems for speed and spatial attention. However, both Fig. 7). This result is in congruence with the earlier data presented
systems bring about their effects not in temporal succession, but in by Robertson et al. (1998) for a different paradigm, and demon-
parallel, already shortly after cue-induced phasic enhancement of strates the adaptive potential of a damaged attentional network.
alertness. Moreover, the switching of the hemi-eld, in which the However, it also raises the question of the clinical implications of
effects are observed for patients as compared to control subjects, such a short-lived effect. After all, at the longest SOA, the patho-
requires an explanation. After all, if a right-hemispheric dominance logical rightward bias reappeared and the neglect pattern was fully
for alertness is assumed, the greatest effects would be expected to re-established.
be observed within the left hemi-eld, as was the case in control It should be taken into account, however, that a core function
participants. of an alertness cue is to initiate an orienting response (Posner,
Taking into account the damage to the right-sided alerting net- 2008). In reality settings, this is typically accomplished by eye
work in neglect patients (indicated by the lack of intrinsic alertness or head movements which support the aim to foveate a salient
modulations), and the fact that a phasic alerting cue exerted effects stimulus that is available afterwards for close examination. In our
in both hemi-elds (although predominantly on the right side), the study, when such a stimulus did not appear within the 650 ms fol-
greater enhancement of sensory effectiveness in the ipsilesional lowing the alertness cue, the neglect pattern was re-established.
eld can be explained in the following way: Phasic alertness might However, this does not imply that, in case visual stimulation is
activate a system supporting processing speed (indicated by sen- indeed provided, the cue-induced benet resulting is necessarily
sory effectiveness in our study) that leads to a global stimulation only shortlived. In our study we only used briey presented stim-
within both hemispheres. When only a single stimulus is present uli. It remains an open empirical issue whether, under conditions
within one hemi-eld, then the intact hemisphere (the left one of enduring sensory stimulation (such as, e.g. the unlimited pre-
in the neglect patients) will always outperform the damaged one, sentation of a visual search display), the alertness-induced effect
giving rise to a right-hemi-eld advantage in single target displays. might be more stable and even remain signicant for the dura-
The simultaneous effect of phasic alertness on the spatial bias tion of any target scene presentation. Furthermore, it should be
points to the additional cue-induced activation of a second system, tested whether, using a systematic alertness training, the effec-
devoted to spatial attentional weighting, which exerts a specically tive period, i.e. the gap between cue and target, without sensory
right-hemispheric effect. This system would become especially rel- stimulation, can be prolonged. Nevertheless, should alerting cues
evant in situations of inter-hemispheric competition, as in the be used as a therapeutic tool in a clinical setting, then the SOA
bilateral target displays of the whole-report paradigm of our study, between an alerting cue and a target should be adjusted to the
and, thus, would lead to a more balanced spatial weighting in these effective short-time window documented in our study in order to
conditions. reach the maximum benet, at least at the beginning of an alertness
The rapidly evolving effects on sensory effectiveness and spa- training.
tial bias following the cue suggest that the two simultaneously In summary, based on the present results, it can be concluded
activated systems mediating speed and spatial weighting rather that higher levels of alertness can help overcome the typical neglect
represent lower-level centres. Therefore, one plausible neural can- symptoms, such as rightward lateralization and unilateral extinc-
didate structure underlying these effects is the thalamus, which tion, indicative of the relevance of alertness in disturbed attentional
is known to be relevant both in the ascending thalamic mes- competition and, thus, spatial-attentional asymmetries. Notably,
encephalic phasic alertness system (Fan, McCandliss, Fossella, a quick redistribution of attentional weights across both hemi-
Flombaum, & Posner, 2005; Posner & Petersen, 1990; Sturm & elds has been identied as the major source of the performance
Willmes, 2001; Thiel & Fink, 2007) and which has also been benet even at the single case level, i.e. in each of the patients
proposed to be critical for visual extinction by several authors tested. Our results are in accordance with previous studies showing
(Fimm et al., 2001; Karnath, Himmelbach, & Rorden, 2002; Rafal a rightward bias in exploration behaviour and a unilateral right-
& Posner, 1987). For instance, Habekost and Rostrup (2006) ward lateralization in neglect (e.g. Bartolomeo & Chokron, 2002;
reported a rightward spatial bias in a comparable TVA-based Bublak et al., 2005; Duncan et al., 1999; e.g. Heilman et al., 2003;
paradigm following stroke in the right-sided pulvinar nucleus of Karnath, 1988). In addition, they provide important clues as to
the thalamus. In fact, on the neural interpretation of the TVA both the time course and the effective mechanism responsible for
(NTVA) proposed by Bundesen, Habekost, & Kyllingsbk (2005), the alleviation of these decits. Due to our low sample size and
the pulvinar nucleus of the thalamus is assumed as the critical the limitations of our neuroanatomical evidence, these clues have
region mediating spatial-attentional weighting across the visual to be further corroborated by future studies with larger sample
eld. sizes also aiming at a closer examination of longer-term training
Based on our results, we would suggest that both thala- effects and a better understanding of the precise neuroanatomical
mic subsystems are simultaneously activated by another system correlates
K. Finke et al. / Neuropsychologia 50 (2012) 11781189 1189