Implicit awareness in anosognosia for hemiplegia: unconscious interference without conscious re-representation

Some patients with anosognosia for hemiplegia, i.e. apparent unawareness of hemiplegia, have been clinically observed to show 'tacit' or 'implicit' awareness of their deficits. Here we have experimentally examined whether implicit and explicit responses to the same deficit-related material can dissociate. Fourteen stroke patients with right hemisphere lesions and contralesional paralysis were tested for implicit and explicit responses to brief sentences with deficit-related themes. These responses were elicited using: (i) a verbal inhibition test in which patients had to inhibit completing each sentence with an automatic response (implicit task) and (ii) a rating procedure in which patients rated the self-relevance of the same sentences (explicit task). A group of anosognosic hemiplegic patients was significantly slower than a control group of aware hemiplegic patients in performing the inhibition task with deficit-related sentences than with other emotionally negative themes (relative to neutral themes). This occurred despite their explicit denial of the self-relevance of the former sentences. Individual patient analysis showed that six of the seven anosognosic patients significantly differed from the control group in this dissociation. Using lesion mapping procedures, we found that the lesions of the anosognosic patients differed from those of the 'aware' controls mainly by involving the anterior parts of the insula, inferior motor areas, basal ganglia structures, limbic structures and deep white matter. In contrast, the anosognosic patient without implicit awareness had more cortical lesions, mostly in frontal areas, including lateral premotor regions, and also in the parietal and occipital lobes. These results provide strong experimental support for a specific dissociation between implicit and explicit awareness of deficits. More generally, the combination of our behavioural and neural findings suggests that an explicit, affectively personalized sensorimotor awareness requires the re-representation of sensorimotor information in the insular cortex, with possible involvement of limbic areas and basal ganglia circuits. The delusional features of anosognosia for hemiplegia can be explained as a failure of this re-representation.