Updating beliefs beyond the here-and-now: the counter-factual self in anosognosia for hemiplegia

Abstract The syndrome of anosognosia for hemiplegia, or the lack of awareness for one’s paralysis following right hemisphere stroke, can provide unique insights into the neurocognitive mechanisms of self-awareness. Yet it remains unclear whether anosognosia for hemiplegia is a modality-specific deficit of sensorimotor monitoring, or whether domain-general processes of attention and belief-updating converge to cause anosognosia for hemiplegia. Using a Bayesian learning framework, we formalized and empirically investigated the hypothesis that failures to update anosognosic beliefs can be explained by abnormalities in the relative uncertainty (i.e. precision) ascribed to prior beliefs versus sensory information in different contexts. We designed a new motor belief-updating task that manipulated both the temporal (prospective and retrospective) and spatial (hemispace most affected by inattention and hemispace less affected by inattention) conditions in which beliefs had to be updated, and we validated its sensitivity to anosognosia for hemiplegia in 26 patients with right hemisphere stroke. We then computed and empirically tested two different Bayesian predictors of prospective beliefs using two proxies for precision in anosognosia for hemiplegia patients: (i) standardized, neuropsychological measures of objective attention abilities, i.e. visuospatial neglect scores and (ii) subjective uncertainty reports, i.e. confidence ratings. Our results suggest that while neglect does not affect local, sensorimotor error monitoring, it does seem to affect the degree to which observed errors are used to update more general, prospective beliefs about counterfactual motor abilities in anosognosia for hemiplegia. Difficulties in such ‘counterfactual’ belief-updating were associated with disruptions in tracts of the ventral attentional network (i.e. superior longitudinal fasciculus connecting the temporo-parietal junction and ventral frontal cortex) and associated lesions to the insula, inferior parietal cortex and superior temporal regions. These results suggest that self-awareness extends beyond local, retrospective monitoring, requiring also salience-based, convergence of beliefs about the self that go beyond the ‘here-and-now’ of sensorimotor experience.

x 181 mm, voxel resolution: 1 mm2) was rotated to match the orientation of the patient's MRI or CT scan. Lesions were outlined on the axial slices of the rotated template. The resulting lesion volumes were then rotated back into the canonical orientation, to align the lesion volumes of each patient to the same stereotaxic space.
Finally, to exclude voxels of lesions outside white and grey matter brain tissue, lesion volumes were filtered by means of custom masks based on the ICBM152 template.
Lesion volume analysis was computed with Non-Parametric Mapping (NPM) software, 8 in order to ensure no difference between groups that could explain lesion and behavioural results.
First, a lesion overlap was calculated to create a color-coded overlay map of injured voxels across patients in each group to provide an overview of all lesioned brain areas in each group (AHP and HP, see Supplementary Fig.4). In order to investigate the neural correlates of AHP, we first compared the lesion maps of the two groups by means of a subtraction technique and a voxel-based lesion comparison based on the Libermeister binomial test (False discovery rate FDR corrected) 9 . The outcome of the subtraction and the significant lesion maps of the voxel-based lesion techniques were superimposed onto T1 templates (AAL) to calculate the number of lesioned voxels in cerebral areas and the centre of the mass in each damaged area.
Statistical contribution of lesion location to deficit in general motor awareness was also tested using voxel-based lesion symptom mapping 10  All lesion results were visualised in MRIcron. Three anatomical templates served to identify grey and white matter region labels: the "automated anatomical labelling" (AAL) template, 11 the JHU white-matter tractography atlas, 12 and the "NatBrainLab" template of the "tractography based Atlas of human brain connections Projection Network". 13,14 Supplementary Results

Importance Ratings
To ensure basic perception and relevance of the experimental materials, the objects corresponding to bimanual everyday actions (e.g. using cutlery) were positioned on a

Confidence Estimates
Confidence estimates were analysed separately at each time point. AHP patients were significantly less confident than HP patient only on the retrospective confidence Error bars represent the standard error of the mean.

Retrospective Updating in AHP and HP patients
To examine the degree to which patients were able to observe that they have failed to perform an action despite their unrealistic prior prospective estimates, we used a 'Retrospective Updating' index i.e. the difference between retrospective and prior prospective performance estimates for each patient.
Directly after an execution attempt, AHP patients updated their performance estimate more than HP patients (Z=2.404, p=.015, ɳ 2 =.222; Fig.3). This can in part be explained by the fact that AHP patients have a higher Prior Prospective performance estimate than HP, and HP patients' Prior and Retrospective estimates are close to 0.
Patients updated their performance estimate after execution similarly on the

Lesion Mapping Supplementary Results
First, an analysis of the volume of the lesions was computed with Non-Parametric Mapping (NPM) software (Rorden et al., 2007) and showed no significant difference between AHP and HP patients (Z=-1.272, p=.217, ɳ 2 =0.062).

Damaged areas related to anosognosia for hemiplegia
In order to investigate the neural correlates of AHP, we first compared the lesion maps of the AHP and HP groups ( Supplementary Fig.4A&B) Fig.4D, and Supplementary Table 1A).
As an illustration, we also computed the subtraction map, showing regions more damaged in AHP patients than in HP patients (>30%; See Supplementary Fig.4C).
VLSM analysis using the inversed continuous Feinberg awareness scores, revealed that voxels within several regions including the amygdala, hippocampus, as well as the insula, the supramarginal, the angular and superior and mid-temporal gyrus (SMG, AG and STG), were associated with differences in awareness (see