Social dysfunction relates to shifts within socioaffective brain systems among Schizophrenia and Alzheimer ’ s disease patients

Social dysfunction represents one of the most common signs of neuropsychiatric disorders, such as Schizophrenia (SZ) and Alzheimer ’ s disease (AD). Perturbed socioaffective neural processing is crucially implicated in SZ/AD and generally linked to social dysfunction. Yet, transdiagnostic properties of social dysfunction and its neuro-biological underpinnings remain unknown. As part of the European PRISM project, we examined whether social dysfunction maps onto shifts within socioaffective brain systems across SZ and AD patients. We probed coupling of social dysfunction with socioaffective neural processing, as indexed by an implicit facial emotional processing fMRI task, across SZ ( N = 46), AD ( N = 40) and two age-matched healthy control (HC) groups ( N = 26 HC-younger and N = 27 HC-older). Behavioural (i


Introduction
The current symptom-based, categorical nosology of neuropsychiatric disorders allows for a pragmatic approach to treatment choice and basic clinical research (Kas et al., 2019).Crucially, however, it fails to address clinical and neurobiological heterogeneity within and across disorders and hinders the development of more effective and personalized treatments.The Research Domain Criteria (RDoC) framework aims to tackle these limitations, by acknowledging the overlap and heterogeneity across neuropsychiatric disorders, and describing them along domains of transdiagnostic phenotypes, rather than separable disease categories (Insel et al., 2010;Kas et al., 2007;Insel and Cuthbert, 2015).To this end, different levels of data are integrated (from observable behaviour to neurobiological measures) to classify mental disorders based on underlying pathophysiology (Insel and Cuthbert, 2015).
One of the promising transdiagnostic domains suggested within RDoC is that of social (dys)function.This domain comprises social communication, perception of self/others, affiliation, and attachment, which seem to be underpinned and influenced by distinct behavioural and neurobiological processes (Insel and Cuthbert, 2015;Porcelli et al., 2019).Social dysfunction is one of the first and most common signs of neuropsychiatric disorders, likely because of the enormous amount and complexity of brain processes required to initiate and maintain adaptive social behaviour (Porcelli et al., 2019;van der Wee et al., 2019).Importantly, behavioural aspects of social dysfunction (i.e.social withdrawal) have been shown to be predictive of worse treatment outcomes across diverse psychiatric disorders (Oliva et al., 2022).Converging lines of research report shared negative symptomatology such as social dysfunction across multiple neuropsychiatric disorders (Winograd-Gurvich et al., 2006;Insel et al., 2010;Porcelli et al., 2019;Ike et al., 2020).This advocates for neurobiological systems that underpin social function that may be distinct and partly independent of the current neuropsychiatric nosologies (Buckholtz and Meyer-Lindenberg, 2012;Porcelli et al., 2019;Saris et al., 2022).Empirical data in support of this hypothesis are, however, largely lacking and this tends to preclude a thorough understanding of social dysfunction as a transdiagnostic phenotype rather than a symptom of neuropsychiatric nosologies (Kas et al., 2007;Insel et al., 2010).
To this end, the European PRISM (Psychiatric Ratings using Intermediate Stratified Markers) project examined the transdiagnostic value of social dysfunction, and its putatively distinct neurobiological correlates in two distinctive disorders: Schizophrenia (SZ) and Alzheimer's disease (AD) (Kas et al., 2019).Although these two disorders differ in core symptoms, genetic profile, and underlying neurobiology, they importantly overlap in social dysfunction (i.e., social withdrawal, interpersonal dysfunction, loneliness), thus providing a strong test of the transdiagnostic hypothesis proposed by the PRISM study (Saris et al., 2022;Cuthbert, 2019;Kas et al., 2019).PRISM recently showed that social dysfunction is indeed transdiagnostically coupled with the intrinsic connectional integrity of canonical brain systems (i.e.Default Mode Network (DMN)) (Saris et al., 2021), as well as with explicit perceptual processing of emotions in SZ/AD patients and healthy controls (HC) (de la Torre-Luque et al., 2021).Furthermore, it has been shown that SZ/AD patients and HC can be clustered into specific social profiles, using digital biomarkers, that are associated with multimodal neurobiological parameters (Kas et al., 2024).
We report here on how social dysfunction in SZ and AD patients, independent of diagnosis (i.e.transdiagnostic), may map onto socioaffective neural processing of various emotions, using an implicit facial emotion processing (iFEP) fMRI task.The iFEP task typically engages fronto-parieto-limbic networks governing socioaffective function, often showing anomalies in SZ and AD patients (van der Wee et al., 2019;Porcelli et al., 2019;Godlewska et al., 2012), as illustrated in Fig. 1.The current study builds on prior PRISM work, wherein resting-state fMRI was employed to examine intrinsic brain network connectivity in relation to social dysfunction, across the same participants (Saris et al., 2021).
Facial emotion processing was specifically targeted, a) as it is deemed critical to processing social information and subsequent interactions, b) is underpinned by well-characterized brain systems, c) and holds reverse-translational potential into animal models for further dissection of underlying neurobiology (van der Wee et al., 2019;Peleh et al., 2019;Porcelli et al., 2019).Diagnostic nosologies were accounted for in all our analyses, while key clinical and demographic variables were corrected for to further aid robustness and specificity of findings.Based on prior work (Fig. 1, (Porcelli et al., 2019)), we hypothesized that during iFEP task performance, social dysfunction would map onto functional shifts within fronto-parieto-limbic networks (including, but not limited to, regions of the DMN, Saris et al., 2021) supporting socioaffective processing (van der Wee et al., 2019;Porcelli et al., 2019).We also anticipated these effects to be distinct and partly independent of diagnosis (Porcelli et al., 2019).

Participants
Data were from the PRISM study and included 46 SZ and 40 CE patients, along with 53 matched HC participants (26 HC younger: 18-45 years & 27 HC older: 50-80 years) (Bilderbeck et al., 2019;Saris et al., 2021;Kas et al., 2019).Participants were recruited between April 2017 and April 2019 from three sites in the Netherlands (University Medical Centre Utrecht, Amsterdam UMC, and Leiden University Medical Centre) and two sites in Spain (Hospital General Universitario Gregorio Maranon & Hospital Universitario De La Princesa) (Saris et al., 2021).Data collection (including MRI) and participant-level assessments were performed according to harmonised protocols across sites (Saris et al., 2021).The sample size was determined by a priori power analysis that is described elsewhere (Bilderbeck et al., 2019).The study was approved by the Ethics Review Board of corresponding countries and by local review boards of all participating centres.All participants provided verbal and written informed consent prior to participation and were considered as sufficiently competent to participate by researchers and caregivers.

Clinical assessment
DSM-IV diagnosis of SZ was confirmed using the Mini-International Neuropsychiatric Interview (MINI), with at least one psychotic episode and maximum of 15-year disease duration since diagnosis (Sheehan et al., 1998).SZ patients were allowed to be on stable antipsychotic/anticholinergic/antidepressant medication dosage for at least eight weeks, and were 18-45 years old.Current states of positive and negative symptoms of schizophrenia were measured using the PANSS (positive and negative syndrome scale) (KayFiszbein and Opler, 1987).SZ patients were excluded if they scored ≥22 on the seven-item positive symptoms subscale of the PANSS to rule out an active psychotic episode possibly hampering adequate study participation (KayFiszbein and Opler, 1987;Bilderbeck et al., 2019;Saris et al., 2021).
Diagnosis of probable AD was established according to the National Institute on Ageing and the Alzheimer's Association criteria (McKhann et al., 2011).AD patients had to additionally score 20-26 (i.e.mild AD symptoms) on the Mini-Mental State Examination -second edition (MMSE-2), and be 50-80 years of age (FolsteinFolstein and McHugh, 1975).AD patients with history of strokes, either based on clinical judgement, medical history or imaging results, were excluded (Saris et al., 2021).AD patients were allowed to be on stable acetylcholinesterase/NMDA receptor antagonist/antidepressant medication dosage for at least eight weeks.Cognitive dysfunction was estimated in AD patients using the Alzheimer's Disease Assessment Scale -Cognitive subscale (ADAS-cog) (RosenMohs and Davis, 1984).
As the patient groups differed significantly in age, we also included two age-matched HC groups to mitigate age effects, while additionally correcting for age and age-squared in all analyses (Saris et al., 2021).HC participants exclusion criteria were any history of psychopathologies (as confirmed by the MINI) or neurological disorders, and usage of psychotropics and central nervous system affecting medication (Saris et al., 2021).More detail regarding participant in-and exclusion criteria is provided in the Supplement.

Indicators of social dysfunction
Consistent with our prior work (Saris et al., 2021), the current study examined both the behavioural aspect and subjective experience of social dysfunction.The behavioural aspect of social dysfunction was indexed with the Social Functioning Scale (SFS) (Birchwood et al., 1990).The SFS multidimensionally examines social withdrawal, interpersonal functioning, prosocial activities, recreational activities, independence competence and independence-performance (Birchwood et al., 1990).The SFS subscales were all reverse-scored so that higher score would indicate more social dysfunction, with its total score being used in the final imaging analyses (Saris et al., 2021;Birchwood et al., 1990).The Jong-Gierveld Loneliness (LON) questionnaire was employed to assess the subjective experience of social dysfunction in the form of feelings of loneliness, with higher total scores indicating more loneliness (De Jong-Gierveld and Kamphuis, 1985).The two questionnaires were moderately correlated (Spearman's r = 0.55, P < 0.0001), suggesting that while having some overlap they capture partly different aspects of social dysfunction.

MRI data acquisition
Imaging data were acquired according to harmonised protocols across sites, including the approach of the participants in and outside the scanner.A Philips Achieva 3T MRI scanner (Leiden University Medical Centre) and Philips Ingenia 3T MRI scanners (University Medical Centre Utrecht, Amsterdam UMC) with a 32-channel head coil were used at the Dutch sites, while a Siemens Prisma 3T MRI scanner with a 64-channel head coil was used at the Spanish site.All MRI assessments for both Spanish recruiting sites were performed on a single MRI scanner (see Supplement for details data acquisition).
To mitigate potential scanner site/type effects, an extensive quality assurance protocol was used.First, MRI acquisition protocols were harmonized, including the approach of participants in and outside the scanner.Second, pilot testing with a 'travelling head' was performed (i.e. imaging a team member with the same imaging protocol at the different scanners in the Netherlands and Spain) and revealed no significant scanner site/type effects.Finally, we corrected for scanner site/ type in all fMRI statistical analyses.

Facial emotional processing fMRI task paradigm
The iFEP-fMRI paradigm (Fig. 2) was used to probe neural underpinnings of socioaffective processing (van der Wee et al., 2019).During the task, 120 emotional faces (sad, happy, fear) are presented in rapid succession and grouped in 12 separate blocks (10 trials per block), while participants have to judge the sex of the depicted face by button-press (left for male, right for female).The order of the blocks is fixed across participants.Each block starts with a fixation cross presented for 30 s, allowing the participants to prepare for the rapid succession of facial stimuli.Each face is presented for only 100 ms with an inter-stimulus interval (ISI) of 2.9 s, during which the face is replaced Fig. 1.Fronto-parieto-limbic regions governing key socioaffective processes, which often show anomalies in SZ and AD patients (Porcelli et al., 2019).The image was adapted with permission from Porcelli et al., Neurosci Biobehav Rev. (2019) (Porcelli et al., 2019).FFA = fusiform face area; STG = Superior temporal gyrus; IFG = inferior frontal gyrus; IPL = inferior parietal lobule; ACC = anterior cingulate cortex; TPJ = temporo-parietal junction; PFC = prefrontal cortex; vLPFC = ventrolateral prefrontal cortex; VTA = ventral tegmental area; NAc = nucleus accumbens; SOS = superior orbital sulcus.
with a fixation cross.The participants can make their response during the ISI.The next trial begins immediately after the ISI fixation cross.At the end of the task, another 30 s fixation cross is presented, bringing the total task duration to ~12 min.Stimuli were presented and responses were recorded using software developed by P1vital Products Limited.
Prior to scanning, participants were familiarized with the experimental task with a practice session outside the scanner, using other stimuli than in the actual iFEP fMRI task.

fMRI data preprocessing
All data were subjected to standard preprocessing steps, using FMRIB Software Library (FSL) version 6.3 (Jenkinson et al., 2012).Preprocessing consisted of nonbrain-tissue removal, motion correction, grand mean-based intensity normalization of the entire 4-D data set by a single scaling factor, slice timing correction, spatial smoothing with a 5 mm full width at half maximum Gaussian kernel, and temporal high-pass filtering at 0.01 Hz (Gaussian-weighted least-squares straight line fitting).Finally, the functional data were registered to T1-weighted anatomical images, and subsequently to the 2-mm MNI standard space image, using boundary-based registration with 12 • -of-freedom and integrated distortion correction.On top of this, independent component analysis (ICA) based automated removal of motion artefacts (AROMA) was used for (micro)motion-related artefact removal.White matter/cerebrospinal fluid signal removal was implemented to further clean the data of noise (Pruim et al., 2015).The maximum allowed mean displacement due to excessive head motion was set at 3 mm translation or 3 • rotation in any direction.In total, MRI data was available for 149 participants, 10 (SZ=4; AD=5; Older HC=1) were excluded because of excessive motion and poor imaging quality, leaving 139 participants (SZ=46; AD=40; Younger HC=26; Older HC=27) in the final imaging analyses.Table 1 provides detailed description of these participants.

fMRI data analyses
Subject-level statistical analyses using FILM (FMRIB's Improved Linear Model) with local autocorrelation correction were performed in FSL/FEAT.This entailed a general linear model (GLM) wherein regressors for each emotion category were convolved with a doublegamma hemodynamic response function.To measure emotion-specific neural responses, the following contrasts were tested in these subjectlevel GLM's: Fear vs. Happy & Sad (1 − 0.5 − 0.5); Happy vs. Fear & Sad (1 − 0.5 − 0.5); Sad vs. Fear & Happy (1 − 0.5 − 0.5).These subjectlevel statistical maps were then fed into a group-level GLM analysis (whole-brain), to probe how social dysfunction may map onto taskrelated socioaffective neural processing.
This was done using non-parametric permutation-based GLM analyses with automatic outlier-deweighting, as implemented within FSL's Randomise tool (5000 permutations).The GLM included the individual participants' SFS (i.e., behavioural social dysfunction) and LON (i.e., subjective experience social dysfunction) total scores as separate regressors, wherein separate contrasts (contrasts of parameter estimates) probed the unique effects of these two constructs on task-related activity.The GLM parameter estimation in FSL automatically removes the effects of shared variability, so that each effect is adjusted for all others (MumfordPoline and Poldrack, 2015; Jenkinson et al., 2012).Entering SFS and LON in the same group-level GLM takes into account their shared variance, thus revealing task-fMRI activity patterns uniquely associated with each construct (MumfordPoline and Poldrack, 2015;Jenkinson et al., 2012).That is, construct-specific variance in task-fMRI activity over and above what can be explained by the other construct.
This dimensional analysis tested whether across the sample any linear associations could be found between task-related neural activity and individual participant's SFS or LON scores.Diagnostic status (i.e., SZ/AD/HC-younger/HC-older) was also entered in this GLM as regressor, in order to correct for it and disentangle impact of diagnostic status on task-related neural activity from that of social dysfunction.Key clinical (psychotropic medication, comorbid symptomology) and sociodemographic (age, age squared, sex, education, scan site/type) factors were corrected for in the analyses to aid robustness and reliability of findings (added as separate regressors in the GLM).Post-hoc analysis additionally assessed whether diagnosis x SFS and diagnosis x LON interactions effects could be identified, and whether these impacted linear brain-behaviour associations we documented across participants.Analyses were performed with all independent variables demeaned across groups, with statistical thresholding and correction for multiple comparisons achieved through Threshold-Free Cluster Enhancement (TFCE) with family-wise error (FWE) correction at P < 0.05 (Smith and Nichols, 2009).

Patient characteristics
Most SZ patients were treated with antipsychotic medication (93.5%), while some were (also) treated with an antidepressant (19.6%), see Table 1.Nearly half (40%) of the AD patients were treated with acetylcholinesterase and/or NMDA receptor antagonist, while some were (also) treated with antidepressant medication (17.5%).In SZ, the mean (± SD) positive and negative symptoms measured with the PANSS were, respectively, 11.0 (± 3.5) and 14.8 (± 6.0), indicating mild psychopathology.Furthermore, the AD group had a mean ADAS-cog score of 27.3 (± 7.6), indicating mild cognitive dysfunction.SZ patients had higher total reversed SFS scores and higher LON scores compared to all other groups (P's<0.001).Furthermore, AD patients had higher total reversed SFS scores compared to the control groups (P < 0.0001), while their LON scores were similar to AD controls (P > 0.05).
No association was found between total SFS scores and brain activity in response to fearful faces across SZ/AD/HC participants (P > 0.05, TFCE & FWE corrected).Of note, the analyses also revealed no associations between socioaffective neural processing and total LON (loneliness) scores across SZ/AD/HC participants (P's>0.05,TFCE & FWE corrected).
Given the correlation between SFS and LON, separate GLM analyses were run for SFS and LON in post-hoc analysis, so to probe possible collinearity effects.Results remained practically the same, however, with the only difference being that SFS effects now became spatially slightly more distributed (see Figure S1).Post-hoc analysis additionally assessed diagnosis x SFS and diagnosis x LON interaction effects.The analysis revealed no significant interactions (P's>0.05),while brainbehaviour associations we documented remained unchanged/significant (SFS -Sadness coupling P = 0.0174 and SFS -Happiness coupling P = 0.0318).For completeness, the accuracy rates and reaction times for the iFEP task are shown in supplementary Table 1, which were not further analysed given the implicit nature of the task.

Diagnosis and socioaffective neural processing
In addition to examination of interdependencies between social dysfunction and socioaffective neural processing, our fMRI GLM model also simultaneously investigated the impact of diagnostic status on socioaffective neural processing (see Methods, FMRI Data Analyses).The analyses revealed lower activity in the cerebellum among patients (SZ and AD combined) compared to HC (HC-younger and HC-older  combined) in response to sad faces (P = 0.0370, TFCE & FWE corrected) (Supplementary Figure 2D).Additionally, higher activation was found in the visual cortex among patients compared to HC in response to fearful faces (P = 0.0458, TFCE & FWE corrected) (Supplementary Figure 2E).Higher activity was found in the cerebellum and visual cortex among SZ patients relative to their HC peers in response to fearful faces (P = 0.0450, TFCE & FWE corrected) (Supplementary Figure 2A).Lower activity was found in the motor cortex and mid-cingulate area in SZ patients relative to AD patients in response to happy faces (P = 0.0296, TFCE & FWE corrected) (Supplementary Figure 2B).In response to sad faces, lower activity was found in the mid/posterior INS and TPJ regions among SZ patients compared to their HC peers (P = 0.0440, TFCE & FWE corrected) (Supplementary Figure 2C).These disorderspecific diagnostic findings (or their lack of) clearly diverge from the observed social dysfunction effects across all participants, being spatially different and directionally distinctive (increased vs. decreased), further corroborating the specificity of social dysfunction effects documented here.

Discussion
This study demonstrates how social dysfunction maps onto socioaffective brain systems across SZ and AD patients and matched HC participants.Specifically, more severe behavioural social dysfunction across SZ/AD/HC participants related to hyperactivity within frontoparieto-limbic brain systems in response to sad emotions (P = 0.0078), along with hypoactivity of these brain systems in response to happy emotions (P = 0.0418).These effects were independent of diagnosis, and not confounded by key clinical (psychotropic medication, comorbid symptomatology) and sociodemographic factors (age, age squared, sex, education, scan site/type).Effects also proved rather specific to behavioural aspects of social dysfunction, as similar relationships were not found for the subjective experience of social dysfunction (i.e., perceived loneliness).These findings pinpoint altered socioaffective neural processing as a putative marker for social dysfunction, and could aid personalized care initiatives grounded in social behaviour.

Social dysfunction relates to altered sad and happy emotion processing
During processing of sad emotional faces, a coupling emerged across SZ/AD/HC participants, wherein more severe behavioural social dysfunction related to hyperactivity within a collection of frontal (mPFC/OFC/dlPFC/ACC), parietal (TPJ/IPL/precuneus), and limbic (INS/amygdala/hippocampus/striatum/brainstem) brain regions.These fronto-parieto-limbic regions are surmised to work in tandem and fulfil specialized functions when processing information carrying socioaffective salience (Porcelli et al., 2019).In this context, the interconnected limbic regions ostensibly serve reflexive salience processing and visceroaffective sensation/responding, with the frontal regions providing subsequent cognitive control and contextualization to instil adaptive selection and appropriate action.The parietal regions, which are also key nodes within the posterior DMN, then kick in to allow for interpersonal colouring of processed information and mentalising (Barrett and Satpute, 2013;Andrews-Hanna et al., 2010;LiMai and Liu, 2014;BickartDickerson and Barrett, 2014).One may thus speculate that sad facial emotions are disproportionally seen as salient socioaffective stimuli, and thus prompt exaggerated neurobiological responses in more socially dysfunctional individuals, regardless of diagnostic status (Porcelli et al., 2019).
Our analyses further revealed that the behavioural aspects of social dysfunction are associated with hypoactivity in the pINS, dlPFC and TPJ in response to happy faces across SZ/AD/HC participants.The pINS has previously been implicated in regulating the physiological reactivity to salient stimuli (Menon and Uddin, 2010).The dlPFC and TPJ have both been implicated in understanding the mental states of others, also known as Theory of Mind (ToM) or mentalising (Porcelli et al., 2019).Moreover, data from the PRISM consortium showed that ToM deficits are associated with social dysfunction across SZ and AD patients (Braak et al., 2022).Taken together, it may be possible that happy facial emotions are perceived as less relevant social sensory cues and evoke a reduced physiological response in individuals with social dysfunction, independent of diagnosis.
However, it remains unclear if social dysfunction gives rise to suboptimal socioaffective neural processing or whether it is the other way around.The cross-sectional nature of this study does not allow for causal inferences, yet tentative empirical data points to shifts in socioaffective brain systems as a putative causal factor.Clinical examples include repetitive TMS and psychotropic medication that manage to normalize functional and connectional integrity of socioaffective brain systems, followed by altered (more favourable) social behaviour (Pievani et al., 2017;Anderson et al., 2016;Lorenzi et al., 2011;Bais et al., 2017;Goveas et al., 2011).Moreover, animal data on stimulation/manipulation of key socioaffective neural systems show stark changes in social behaviour (Ike et al., 2020;Shemesh et al., 2016;Filiano et al., 2016;Mills et al., 2016;Missault et al., 2019;Zott et al., 2018).Additionally, top hits from a recently performed GWAS of the sociability trait were most heavily expressed in some of the frontal-parieto-limbic regions documented here, with sociability and loneliness being genetically linked (Bralten et al., 2019).
So, regardless of its origins, the neurobiological shifts documented here appear to influence a wide variety of core deficits in social, cognitive, and affective functions, which are likely to trigger (sub)clinical symptomatalogies (Porcelli et al., 2019).Future research is needed to generalize and confirm the findings documented here in SZ and AD patients, by including other neuropsychiatric disorders characterized by social dysfunction, so to ultimately accelerate the development of novel personalized treatments for social dysfunction.For instance, patients with affective disorders, such as unipolar and bipolar depression, also present with social dysfunction and exhibit alterations in facial emotion recognition and emotion regulation (De Prisco et al., 2023;Fares-Otero et al., 2023, 2022).Therefore, the newer PRISM2 project also aims to explore the neurobiological correlates of social dysfunction in major depressive disorder (MDD), in addition to schizophrenia and Alzheimer's disease, in the near future.

Social dysfunction not related to altered fear processing
We did not find an association between social dysfunction and altered brain activity in response to fearful faces.This may suggest that the potential negative emotional bias associated with social dysfunction might be more specific to increased sensitivity to sad emotions compared to a more general negative bias that includes fearful emotions (Porcelli et al., 2019).This assumption might be addressed in future studies by probing other negative emotions, such as anger and disgust.Additionally, a recent meta-analysis suggested that the neural processing of emotional faces is organized in latent groupings based on threat content, in which fear and anger have a higher level of threat content compared to happy and sad emotional faces (Lukito et al., 2023).This further indicates that threat perception in emotional faces might be a less important factor for social dysfunction.Lastly, we found no association between the more subjective experiences of social dysfunction (i.e. perceived loneliness) and socioaffective neural processing.In line with this, data from the PRISM consortium showed that perceived loneliness was not associated with facial emotion recognition capacity (de la Torre-Luque et al., 2021;Braak et al., 2022).Some studies have found a link between (socio)affective neural processing and loneliness in SZ, MDD and HC, though not in a transdiagnostic fashion (Lindner et al., 2014;Wong et al., 2016Wong et al., , 2019)).It could be speculated that facial emotion processing as implemented here does not entail transdiagnostic properties that map onto perceived loneliness, at least not in this dataset.Clearly, additional research in larger and more diverse neuropsychiatric populations is needed to further dissect this proposal, which is one of the current objectives for the PRISM2 consortium.

Specificity of social dysfunction effects vs. diagnostic effects
Though this study was not designed to replicate previously well described changes in socioaffective neural processing as a function of SZ or AD diagnosis (Porcelli et al., 2019), we did include diagnostic categories in the model to pinpoint the unique transdiagnostic effect of social dysfunction (Porcelli et al., 2019).This mainly served transparency and sensitivity testing, and was not set up to examine the disorder-specific socioaffective neural correlates of SZ/AD (described extensively elsewhere (Porcelli et al., 2019)).The disorder-specific findings clearly diverge from the social dysfunction effects, being spatially different and directionally distinctive (increased vs. decreased), further corroborating the specificity of social dysfunction effects documented here.So, while patient-status may affect socioaffective function, it importantly does not account for above-mentioned links between social dysfunction and socioaffective brain systems.In fact, our findings robustly showcase that shifts in socioaffective brain systems are a neurobiological correlate of social dysfunction, which are distinct and independent of current neuropsychiatric nosologies.The findings documented here thus support the ongoing paradigm shift from the traditional nosological perspective on neuropsychiatry towards a more transdiagnostic approach of key functional domains and their neurobiobehavioural underpinnings, such as social (dys)function (Insel et al., 2010;Buckholtz and Meyer-Lindenberg, 2012).

Limitations and strengths
Our study has a number of limitations that need to be acknowledged.First, we examined the link between socioaffective neural processing and social dysfunction, as measured independently by two different questionnaires.The use of questionnaires to capture the notoriously complex phenomenon of social dysfunction is a vast simplification.However, to date it is the best proxy available for easily accessible and reliable assessments until more sophisticated techniques are employed in this population (Eskes et al., 2016;Jongs et al., 2020).While we included patients with mild disorder specific symptomatology (i.e., minor cognitive dysfunction and few positive and negative symptoms) and a relatively recent disease onset, we cannot fully rule out the long-term consequences of psychopathology or neurodegeneration on social dysfunction.Additionally, lifetime regular substance use (e.g.cannabis) has been associated with increased facial emotion recognition accuracy in SZ and HC (L Fusar-Poli et al., 2022a).However, drug abuse/dependence within the previous 3 years was an exclusion criteria in the current study (see supplement).Although not very likely, we cannot completely rule out that lifetime regular substance use have had an effect on the association between aberrant socioaffective neural processing and social dysfunction.Finally, distinct neuropathological mechanisms may very well underlie the coupling we documented between social dysfunction and socioaffective neural processing across AD and SZ patients.While impaired socioaffective neural processing in AD might be due to atrophy and neurodegeneration, in SZ it may be caused by neurodevelopmental anomalies.Future studies might address these issues, by using a longitudinal approach and combining different MRI modalities (i.e., activity/connectivity/morphology).
In spite of these limitations, the study certainly adds to our understanding of social dysfunction and its putative neurobiology (Insel et al., 2010;Kas et al., 2019).The study moreover has an innovative approach, combining two distinctive neuropsychiatric disorders with differing disease characteristics on the important topic of social dysfunction.Current findings on social dysfunction may therefore aid data-driven patient stratification initiatives given their transdiagnostic dimensional feature.Similar to previous research, the term 'transdiagnostic' was used in the current study to indicate brain-behaviour relations that can be observed across two or more neuropsychiatric nosologies and HC participants (Porcelli et al., 2019;Saris et al., 2021;Sheffield et al., 2017;Kebets et al., 2019).Currently, established objective (e.g.neuroimaging) biomarkers are lacking in precision psychiatry (P Fusar-Poli et al., 2022b).Our findings could contribute to the advancement of precision psychiatry, by informing the development of more robust biomarkers and effective treatment strategies that are rooted in social behaviour, and its neurobehavioral underpinnings (Insel et al., 2010;Kas et al., 2019).For instance, more robust biomarkers that are independent of diagnosis may eventually be used for treatment stratification, which can lead to more effective treatment outcomes and improved cost effectiveness, and for individualized prognostic purposes (P Fusar-Poli et al., 2022b).Ideally, future studies should further explore and validate the findings reported here, and check its generalizability in additional neuropsychiatric disorders (depression/anxiety).

Conclusion
In sum, our findings suggest that social dysfunction relates to shifts within socioaffective brain systems across SZ/AD patients and HC.These findings pinpoint altered socioaffective neural processing as a putative marker for social dysfunction, and could aid personalized care initiatives grounded in social behaviour.These initial exploratory findings should be further validated, for example through multimodal examination of socioaffective brain systems and complex network analyses, to attain a more fine-grained understanding of their contributions to social (dys) function.

Contributors
All authors contributed to this study.SB, DN and MA wrote the original manuscript.SB and MA performed the formal analysis.SB, BP, TS, YP, AVC, ATL and MA were involved in the conceptualization of the research.AVC, IMJS, LMR and JAL conducted the study protocols.BP, YP, ACB, AM, GRD, HM, JLA, CA, NW, MJK and MA were involved in study supervision.All the authors were involved in review and editing the final manuscript.

Fig. 2 .
Fig. 2. Time course of stimulus presentation for the implicit facial emotional processing task during the scanning session.

Fig. 3 .
Fig. 3. Social dysfunction and sad emotion processing.The left panel depicts fronto-parieto-limbic brain regions wherein more behavioural social dysfunction (SFS scores) was associated with stronger neural responses to sad faces.Both cortical and subcortical brain surfaces are shown for better visibility of effects.The scatter plots provide a quantitative visualization of this effect, wherein mean neural response to sad emotions across all illuminated voxels (y axis) are plotted against behavioural social dysfunction levels (x axis).The values on y and x axis are Z-score residuals.The grey solid line depicts the slope of the association, with the dotted bands indicating the 95% confidence interval of the slope.Higher positive values on the x axis indicate more severe social dysfunction.The yellow-orange scalar bar reflects significance level (P-value, TFCE & FWE corrected) of this brain-behaviour association across all participants.

Fig. 4 .
Fig. 4. Social dysfunction and happy emotion processing.The left panel depicts lateral views of frontal, insular, and temporoparietal brain regions wherein more behavioural social dysfunction (SFS scores) was associated with weaker neural responses to happy faces.The scatter plots provide a quantitative visualization of this effect, wherein mean neural response to happy emotions across all illuminated voxels (y axis) are plotted against behavioural social dysfunction levels (x axis).The values on y and x axis are Z-score residuals.The grey solid line depicts the slope of the association, with the dotted bands indicating the 95% confidence interval of the slope.Higher positive values on the x axis indicate more severe social dysfunction.The yellow-orange scalar bar reflects significance level (P-value, TFCE & FWE corrected) of this brain-behaviour association across all participants.

Table 1
Sample characteristics of each study group.
Mean and standard deviations (SD) are displayed for continuous variables.When assumptions were violated, median and Q1 -Q3 are also displayed.Chi-square tests were performed for categorical variables.For continuous variables, independent sample t-tests were performed and when assumptions were violated, Mann-Whitney U tests were performed.Alzheimer's severity ADAS-COG scores are based on N = 39.PANSS = Positive and Negative Syndrome Scale.ADAS-COG = Alzheimer's Disease Assessment Scale -Cognitive subscale.SFS = Social Functioning Scale.