In our study of gesticulation in medication-naïve ARMS subjects, less deictic gestures in ARMS were observed compared with controls. This gesture decrease was also correlated with more negative and positive symptoms. Besides, negative symptoms correlated with metaphoric, and positive symptoms with beats gestures. At last, less gesticulation was significantly correlated with reduced total amount of movement and higher variability of movement.
Our study is in line with the previous one by Mittal et al. (2006), who reported reduced gestures rates in 20 SPD individuals13. Here, ARMS and controls differed only for deictic gestures and this difference might be related to sample differences. For instance, age impacts on gestural performance—their participants were aged 12 to 18 while ours were between 18 and 25 years37. Also, our study analyzed these gestures categories in a Brazilian sample of ARMS individuals, culture also mediating gesture expression37–39. In addition, methodological differences may also have influenced these results: while our analysis was performed using two different free speech tasks, they filmed the first 30 minutes of a structured diagnostic interview. Our objective was to use tasks that allowed the assessment of gesticulation in a closer way to those performed in spontaneous interactions. Aiming for more naturalistic data, we tried, thus, to reduce the tension that participants may feel during diagnostic interviews that could end up accentuating gestures differences between ARMS and control subjects.
Similar to our findings, poor gesture performance is also observed in schizophrenia. In addition to speaking less, 20 patients diagnosed with schizophrenia also made less use of hand gestures during interaction with healthy controls to discuss a moral dilemma4. Reduced gestures were also seen for 25 drug-free schizophrenia patients in an ethological study40. Impairments in imitation and pantomime hand gestures are pointed out in several studies41–46. For example, using the Test of Upper Limb Apraxia (TULIA) to assess these categories in 30 schizophrenia patients, it was observed that 66.7% presented deficits41. However, increasing this sample for 89 patients only 52% presented gestures impairments8. Similarly, from a sample composed by 40 patients with schizophrenia and 41 healthy controls in Switzerland, 19 patients presented impairments47.
It is possible that these differences found in schizophrenia studies are related to deficits' evolution during the course of the disease. For instance, a study conducted by Stegmayer et al. (2016) showed that 14 multiple-episode patients presented severe gestures deficits compared to 14 first-episode ones5. On the other hand, Lavelle et al. (2014) interestingly did not find significant differences between schizophrenia patients and psychiatrists regarding gestural movements. Authors pointed to the possibility that these patients might be divided into two different categories of non-verbal behavior: one prosocial to invite interaction and one to avoid social interaction48. Given that here we are investigating deficits in ARMS individuals, it is possible that the lack of differences for other gestures categories might be related to the actual stage that they are or, moreover, that they are part of the group with more prosocial behavior. However, this is speculative and it is necessary to investigate these variables longitudinally, after transition to psychosis.
Regarding the findings for correlation between gestures and SIPS symptoms, there are few studies that evaluate it in ARMS condition. Mittal et al. (2006) found a negative correlation between iconic gestures and positive symptoms, but no correlations for other gestures or for negative symptoms13. Osborne et al. (2017) also did not find any association between beat gestures and SIPS’ positive or negative symptoms14. Although our results are different from theirs, they agree with findings for schizophrenia patients where gesture deficits are mostly associated with negative symptoms4,5,8,12. In schizophrenia gesture performance was also associated with the positive symptom dimension46. Walther et al. (2015) found a negative association between gesture performance and positive symptoms in a study conducted with 46 patients with schizophrenia, schizoaffective or schizophreniform disorder and 44 healthy controls7. We also found a negative correlation between total negative symptoms and some categories (deictic and metaphoric), and between total positive symptoms and deictic and beats gestures. Also, deictic gestures showed to be a significant estimate parameter for negative and disorganization symptoms, and metaphoric gestures for negative and general symptoms. In summary, our findings present convergent validity, in the sense that our behavioral measures are associated with the existing ground truth of clinical ratings of emotional expression.
Gesture deficits in schizophrenia also showed to be an important predictive factor for disease outcome. In a 6-month follow-up study, schizophrenia patients with gestural impairments in baseline showed a higher level of overall negative symptoms after the period when compared to those without such impairments9. Gesturing involves different processing modalities and the coordination of different brain areas13,49. In general, the impairments in schizophrenia are linked to frontal lobe dysfunctions, motor abnormalities and working memory deficits7,42,50. For example, in patients with impairments in gesture production a reduced cortical thickness in eight ROIs, including precentral gyrus, insula, inferior and superior parietal lobe47. However, studies also showed that different gestural categories might be more related to specific regions49–51. So, it is possible that the observable gesture impairments seen in consonance with SIPS symptoms represent early brain dysfunction in ARMS. Future studies should further explore this hypothesis.
At last, we observed that gesture impairments were related to differences in energy of movement—assessed in a previous work of ours. In these findings, a reduced movement and increased variability of movements in ARMS was observed with MEA20. Given that MEA only measures how much the person moved and not the type of movement made, in the present study we were able to see that gesticulation was positively correlated with amount of movement and negatively correlated with movement variability. Thus, the reduced movement obtained with MEA is largely attributed to the decrease in deictic gestures here observed, addressing the gap that we described in the previous study. In addition, these findings show the importance of both manual and automated analysis, which are complementary to each other in the analysis of NVC in ARMS. It is important to highlight, however, that other movement variables beyond the gesture itself, such as postural sway52,53, might also be contributing to the significant differences found in MEA, but no other specific movement parameter was obtained to verify this hypothesis.
Our study has some limitations. First, we worked with a limited sample size. However, our sample was formed by non-helping-seeking medication-naïve individuals and findings were consistent with previously published studies in ARMS54,55. Also, no differences in demographic data was seen between them and the control group, increasing its reliability. Second, a large number of statistical tests were made. We tried to minimize type I error by using the General Linear Model to infer individual symptoms from gestures, but the possibility of statistical bias cannot be ruled out. Third, gesticulation was assessed manually. Considering that we have already used MEA to analyze the total amount of movement in this sample, it was necessary to understand the type of movement performed. For this, the evaluation of different gestural categories depended on the analysis of the semantic context of the verbal component associated with the co-speech gesture and automatic analysis does not currently allow this classification. Thus, this limitation is intrinsic to the analysis performed.
Summarizing, the results we were able to show here are important features to be considered in the visual phenotyping of at-risk individuals. Our findings raise several possibilities, such as the use of gestures analysis in videos collected by naturalistic ways to improve ARMS detection, and the classification of symptom severity in ARMS individuals. As stated before, our results also bring important data that should be used for cross-cultural analysis between different ARMS samples, to assess culture-specific as well as universal parameters on gestural impairments across the psychosis continuum. Future directions point us to the assessment of gestures evolution in these at-risk subjects over time, following up with them to see if any baseline gestural feature has predictive power on outcome, for instance. Also, multimodal analysis—e.g. brain imaging—should be planned to test the use of gesture impairment as a potential endophenotype for schizophrenia spectrum disorders.