The absence of self-awareness is a crucial aspect in the symptomatology of various neurodegenerative disorders. This characteristic becomes relevant due to the strong implications it has on the patient’s quality of life, on the effects that functional dependence has on the caregiver and on the efficacy of the therapy. Faced with a construct as complex as self-awareness, there are in the literature investigations on different aspects of this phenomenon, such as the creation of cognitive models, the study of the neural substrate and the research of appropriate assessment methods that can reliably detect this function. With regard to the assessment methods, there are methodologies in the literature that provide complementary information. The first modality is a quantitatively online measurement based on the discrepancy between the estimate of the patient of his performance and his actual performance, but often neglecting the ecological validity and the real functioning of the subject. The second kind collecting subjective information on the actual daily functioning of the patient resulting from clinical observation or interviews with the subject and caregivers, but obtaining offline information on the functioning of the subject, liable to bias that may imply an overestimation or underestimation of subject’s ability. The absence of acknowledged metacognitive functional assessment with normative data to evaluate awareness winks at the emerging and increasingly consistent use of virtual reality (VR) also in the context of cognitive research and clinical assessment. This article aims to make a theoretical proposal regarding the use of this innovative and promising tool as a supplement to the assessment methods of self-awareness.

In addition to established body image alterations, abnormal perception and executive functioning in anorexia nervosa (AN), neurocognitive factors including multisensory integration (MSI) and episodic memory (EM) might play a pivotal role in the diagnosis of this disorder. According to the allocentric lock theory, deficits in the updating of distorted memory of body-based episodes through misleading real-time multisensory bodily stimuli could lead to altered body image in AN. In this study, 25 healthy females and nine AN individuals were tested on a set of neurocognitive measures, encompassing unimodal perceptual accuracy and MSI (MSI) ability assessed with the sound-induced flash illusion (SiFI), episodic memory recognition (EMR) evaluated with a remember/know (R/K) task, memory and executive functions tested with the RBMT-3 and the Stroop task. Collected data were analyzed with Bayesian statistics and machine learning algorithms. In the SiFI task, we found that AN compared to control group had lower discrimination accuracy for unimodal visual and auditory stimuli, for bimodal (visual and auditory) stimuli and disrupted MSI ability. Further, we found on the EMR task and the RBMT-3 that AN individuals had higher proportions of false memories for R responses and visual recognition. Additionally, we found greater inhibition at the Stroop task for the patient group compared. The importance of the considered neurocognitive measures was confirmed by a machine learning feasibility analysis, which showed that SiFI, RBMT-3, Stroop and EMR had more weight than classic eating disorder risk scales of the EDI-3 when computing classification between the AN and control individuals. In conclusion, MSI along with memory could be crucial factors for improving diagnosis and consequently design innovative therapeutic solutions that tap critical bodily and cognitive elements altered in AN with new technologies such as virtual reality.