Characterizing social motivation deficits in treatment‐resistant schizophrenia: Evidence from social incentive performance Authors: Simon SY Lui, Jasmine WS Chan, Jason LF Chan, Kimi HY Lam, Lok‐Yin Choi, Raisie WK Wong, Perry BM Leung, Na Zhan, Jason WY Wong, Mindi WY Chiu, Jenny PH Lam, Raymond CK Chan Published: 2026 ABSTRACT AbstractAimThe Treatment‐Response‐and‐Resistance‐in‐Psychosis (TRRIP) criteria recognizes negative symptoms as defining features for treatment‐resistant schizophrenia (TRS). Social motivation refers to the desire to pursue social reward (and to avoid social punishment). Impaired social motivation is related to negative symptoms but has seldom been studied in TRS patients. We aimed to characterize social motivation processing in TRS patients and examine the impacts of negative symptoms on social motivation processing.MethodsWe recruited 60 TRS patients, 60 remitted schizophrenia patients and 60 controls. In the Social Incentive Delay (SID) task, participants had to hit the targets within the response window to seek positive/neutral/negative social outcomes in emoji formats; and they also rated their emotions during the anticipation and receipt of social outcomes. Generalized‐estimation‐equation (GEE) models examined the impacts of TRS/remitted schizophrenia, negative symptoms, and the interaction between clinical‐group‐status and cue‐valence on the SID performance, that is, hit‐or‐not, anticipatory pleasure, and consummatory pleasure.ResultsThe interaction between TRS status and social reward/punishment significantly predicted anticipatory pleasure, implicating that TRS patients' anticipatory pleasure was less subject to social reward/punishment compared to controls. Moreover, the interaction between TRS and social reward significantly predicted consummatory pleasure, implicating that TRS patients' consummatory pleasure was less subject to social reward compared to controls. Negative symptoms significantly predicted the SID accuracy in clinical participants, supporting its relationship with social motivation processing.ConclusionTRS patients have altered social motivation processing, affecting anticipatory and consummatory pleasure. Social motivation deficit may be an intervention target for negative symptoms and social functioning impairments in TRS patients.Keywords: negative symptoms, second‐generation scales, social motivation, treatment‐resistant schizophrenia Treatment‐resistant schizophrenia (TRS) is characterized by failure to respond to at least two non‐clozapine antipsychotic medications, persistent psychotic symptoms, and functional deterioration. 1 , 2 The important role of negative symptoms in TRS patients has gained growing attention. On the one hand, the Treatment Response and Resistance in Psychosis (TRRIP) criteria have now recognized negative symptoms as one of the defining features for treatment‐resistance. 1 On the other hand, many TRS patients respond partially to clozapine and suffer from negative symptoms, severely affecting their social functioning. 3 The ‘motivation and pleasure’ (MAP) and ‘expressivity’ (EXP) domains of negative symptoms both cover social and non‐social aspects. 3 , 4 , 5 A recent study measured negative symptoms in TRS patients using the Clinical Assessment Interview for Negative Symptoms (CAINS), 6 and reported that TRS patients having higher levels of CAINS‐MAP and EXP symptoms were more impaired in social functioning than their TRS counterparts with lower levels of these symptoms. 7 Moreover, another recent study 8 measured the subdomains of ‘diminished expressivity’ and ‘social anhedonia’ of negative symptoms in TRS patients using the Positive and Negative Syndrome Scales (PANSS), 9 and reported that only the ‘social anhedonia’ subdomain predicted patients' social functioning. However, symptom rating scales such as the CAINS and the PANSS cannot tap into the cognitive mechanisms for social anhedonia in TRS patients. In psychology, social motivation refers to the desire to pursue and maintain social relationships 10 (which is one of the core psychological needs of humankind), 11 and covers both ‘approach’ motivation (i.e. social drive for seeking positive social reward such as praise) and ‘avoidance’ motivation (i.e. social drive for avoid negative social reward such as blame/rejection). 12 As such, social motivation is considered an extrinsic, autonomous motivation according to the self‐determination theory. 11 The construct of social motivation differs from social cognition, which is defined as the ability to process social‐salient information to understand and interact with the social world. 13 Empirical research using experimental/behavioral paradigms suggests that schizophrenia patients have impaired social motivation. 14 , 15 For instance, schizophrenia patients anticipate less pleasure from positive social outcomes, and expend less effort to pursue positive social outcomes (or to avoid negative social outcomes) than healthy people. 14 Moreover, a recent study administered an effort‐based decision‐making paradigm to schizophrenia patients, and found that they were less willing to choose high‐demand but high‐reward social scenarios than healthy people, suggesting a diminished motivation to positive social reward. 15 To our knowledge, no study to‐date has been conducted to tap into social reward processing and motivation in the TRS population; and it remains unclear whether social motivation processing in TRS patients would differ from remitted schizophrenia patients. Moreover, further research is warranted to investigate whether negative symptoms in TRS patients would affect their social motivation processing. These unresolved issues constitute barriers to development of interventions targeting at social motivation for TRS patients. 3 To address these knowledge gaps, we conducted this comparative study. We aimed to examine social motivation processing in TRS patients, relative to remitted schizophrenia and healthy people. We also aimed to examine the impacts of negative symptoms on social motivation processing in TRS patients. Given that TRS patients usually have difficulty to carry out complicated motivational tasks (such as the effort‐based decision‐making task using social scenarios), 15 we chose a relatively simple experimental paradigm, that is, the Social Incentive Delay (SID) task. 16 , 17 The SID task taps into patients' motivational processing to approach positive social outcomes (i.e. social approach motivation) and to avoid negative social outcomes (i.e. social avoidance motivation). The SID task is a variant of the classic Monetary Incentive Delay (MID) task, 17 , 18 which has been extensively used in schizophrenia samples to tap into motivation processing. 19 , 20 , 21 This study made three hypotheses. First, we hypothesized that TRS patients would differ in social motivation processing from remitted schizophrenia patients and healthy controls. Second, we hypothesized that negative symptoms would have impacts on social motivation processing in both TRS and remitted schizophrenia patients. Third, we hypothesized that TRS patients would have altered social motivation processing relative to remitted schizophrenia patients and healthy controls, even after controlling for the effects of negative and other symptoms. Methods Participants We recruited 60 TRS patients and 60 remitted schizophrenia patients from the early psychosis intervention clinics and step‐down clinics at Castle Peak Hospital (CPH), Hong Kong. Our schizophrenia sample had shorter illness duration than the chronic samples (>10 years of illness) recruited in prior research on negative symptoms in TRS patients. 7 , 8 The inclusion criteria for clinical patients were: (i) DSM‐IV schizophrenia 22 ; (ii) aged 16–55; (iii) ethnic Chinese; and (iv) able to speak and read Chinese. Moreover, TRS participants must have received clozapine, or have non‐response to at least two different non‐clozapine antipsychotics despite adequate duration (>6 weeks) and dosage (>600 mg/day chlorpromazine equivalence). 1 Nevertheless, we did not stipulate that our participants with TRS should have at least 12‐weeks persistent psychotic symptoms, >80% medication adherence, and at least moderate levels of functional impairments according to the TRRIP criteria. 1 For remitted schizophrenia participants, we followed the symptom severity domain of the Andreasen's remission criteria, 23 that is, (i) having scored ≤3 in the P1, P2, P3, N1, N4, N6, G5 and G9 items of the Positive and Negative Syndrome Scale (PANSS). 9 Given our cross‐sectional design, we did not stipulate the duration domain (6 consecutive months) of the Andreasen's remission criteria in our clinical comparison group. 23 The exclusion criteria included: (i) an IQ of <70; (ii) a history of head injury resulting in >30 min of loss of consciousness; (iii) neurological disorders; (iv) substance abuse in the past 12 months; (v) electroconvulsive therapy in the past 6 months; and (vi) uncorrected hearing or visual impairments. Moreover, we recruited healthy controls from nearby community, using the same exclusion criteria applicable to clinical participants. Healthy controls received a structured interview by qualified psychiatrists to ensure the absence of personal and family history of schizophrenia spectrum disorder. This study was approved by the Ethics Committee of New Territories West Cluster (Approval number: PAED‐2023‐044), and carried out in accordance with the World Medical Association Declaration of Helsinki. All participants provided written informed consent. After the assessments, participants were given a supermarket coupon valued HK$50 (approximately US$6) as subjective incentive, regardless of SID performance. Clinical assessments Clinical participants' diagnosis of schizophrenia was ascertained using the Structured Interview for DSM‐IV (SCID‐I). 24 We gathered their demographics and clinical variables (e.g. date of service entry, psychosis onset, number of previous psychiatric admissions, and the current type and dose of psychotropic medications) in the structured interviews, supplemented by reviews of medical records. The antipsychotic medication dosage was converted to daily defined dose (DDD) of olanzapine equivalence. 25 We also estimated the anticholinergic medication burden, 26 based on the current antipsychotic medications at the time of assessments. Participants' IQ was estimated using the Test of Nonverbal Intelligence‐ Fourth Edition (TONI‐4). 27 Besides the PANSS, 9 we administered the CAINS 6 , 28 , 29 to evaluate clinical symptoms. The CAINS is a 13‐item second‐generation scale, consisting of the motivation and pleasure subscale (MAP) and the expressivity subscale (EXP). In addition, we measured parkinsonism features using the Extrapyramidal Side‐effect Rating Scale (ESRS) 30 and the Abnormal Involuntary Movement Scale (AIMS). 31 We also measured depressive symptoms using the 17‐item Hamilton Rating Scale for Depression (HAM‐D). 32 The Drug Attitude Inventory (DAI) 33 and the Medication Adherence Rating Scale (MARS) 34 were used to measure medication adherence. Lastly, the Social and Occupational Functioning Assessment Scale (SOFAS) 35 was used to measure social functioning in schizophrenia participants and controls. The SID Task The SID task 16 , 17 was used to measures social motivation. Details of the SID task has been described elsewhere. 17 Unlike the MID task which uses monetary gain or loss as outcomes, the SID task uses cartoon‐based facial expressions with body gestures (emoji) as social outcomes (e.g. praise or rejection). In brief, the SID task contained 60 trials. Each trial comprised the following phases, namely: (i) cue presentation; (ii) reward anticipation (and delay); (iii) target presentation; (iv) response window and (approach/avoidance) action; (v) feedback; and (vi) outcome phases. In each trial, participants first viewed a cue (either positive, neutral or negative) which indicated three potential social outcomes (reward/praise, neutral or punishment/ rejection). Then, the task had a short delay, during which reward anticipation occurred in participants. After the delay, a target appeared on the screen, and participants had to press the button as soon as possible to hit the target, within a personalized ‘response window’ (see below). Immediate feedback regarding hit or miss would be given, followed by outcome presentation, i.e. cartoon‐pictures of facial expressions (praise/neutral/rejection) and body gestures (e.g. thumbs‐up/thumbs‐down/no‐gesture) in the outcome phase of the trial. During the reward anticipation phase and the outcome phase, participants were asked to self‐rate their anticipatory and consummatory pleasure on a 9‐point Likert scale shown on the computer screen. In the SID task, each condition (positive, neutral, or negative cue) contained 20 trials, which were presented sequentially in a pseudo‐random manner. The paradigm is illustrated in Fig. 1. Diagrammatic illustration of the Social Incentive Delay (SID) task. In each trial, a cue indicating the condition (a triangle/square/circle) would be followed by a delayed period (a white ‘x’). Then a target (a blue fixation cross) would appear on the screen, and participants have to press a button as quickly as possible. A personalized response window applies to each participant, based on the results of a simple reaction time task, such as the mean hit rate in the neutral‐cue condition is set as 66%. Based on the participant's response (hit or missed) and the cue conditions, the computer shows the cartoon‐based social reward or punishment. 反應夠快 = hit; 反應不夠快 = missed. Before the formal SID task, we administered a simple reaction time task to participants and estimated each participant's individual reaction time. We then adjusted the ‘response window’ for neutral targets on a participant‐by‐participant basis, such that the hit rate for the neutral condition was set as approximately 66%. Because of the personalized response window for the neutral condition, the hit rate of the SID task is regarded as ‘neutral‐cue‐centered’. However, the hit rate for reward and punishment conditions could vary differently among participants. The computerized SID task recorded the hit‐or‐not response, reaction time (in ms), anticipatory pleasure and consummatory pleasure ratings, the number of trials (ranged from 1 to 60), and cue/reward condition for each trial. If the participant missed the target, the reaction time would be arbitrarily recorded as 0. Data analysis Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) version 21.0 for Windows, and the software R version 4.0.3. The significance was set at P < 0.05 (two‐tailed) unless otherwise specified. ANOVAs were used to compare the three groups in age, education, non‐verbal IQ and HAM‐D. We then compared TRS patients and remitted schizophrenia patients in clinical symptoms, number of previous psychiatric admissions, antipsychotic medications, anti‐cholinergic burden, mediation adherence, extra‐pyramidal side‐effects and SOFAS, using independent sample t‐tests. Although previous research on SID used mixed ANOVA models for data analysis, 17 this traditional method had limitations, such as violation of normality assumption, and potential confounds of practice effect as number of SID trials increased. In this study, we conducted generalized estimation equation (GEE) models 36 to address the abovementioned limitations. The GEE models examined the impact of the clinical status of TRS and remitted schizophrenia, as well as the social reward/punishment on the SID performances (i.e. hit‐or‐not, anticipatory pleasure and consummatory pleasure). Specifically, we entered the binary variables of positive cue (social reward), negative cue (social punishment), TRS, remitted schizophrenia as the independent variables, and hit‐or‐not as the outcome variable in Model 1a, with the subject number as the cluster variable, and sex, age, non‐verbal IQ, anticholinergic burden, antipsychotic dosage (DDD), and trial number as covariates. Moreover, to examine the interaction between group and valence of cue, we computed four additional independent variables, i.e. TRS × social reward, TRS × social punishment, remitted schizophrenia × social reward, and remitted schizophrenia × social punishment, and entered these four variables into the same GEE model to form Model 1b. Likewise, Models 2a and 2b were built using the same set of independent variables, cluster variable and covariates, except that the outcome variable became anticipatory pleasure. Lastly, in Models 3a and 3b, consummatory pleasure was entered as the outcome variable. All the GEE models used exchangeable correlation structure. We used binomial distribution to model hit‐or‐not; and Poisson distribution to model anticipatory and consummatory pleasure. We did not build any GEE model for reaction time, because non‐hit trials lacked accurate estimates for the variable. To account for the effects of negative and other clinical symptoms on social motivation processing, we modified the above six GEE models and only used our clinical participants as the sample. Specifically, we computed the binary variable of TRS status by taking the remitted schizophrenia group status as the reference and included the PANSS‐positive symptoms, PANSS‐general symptoms, CAINS‐MAP, and CAIN‐EXP as the additional covariates. To account for the effects of active psychiatric symptoms of TRS patients on social motivation processing, we categorized our TRS participants into the ‘remitted‐TRS’ group and the ‘non‐remitted’ TRS group. 23 Specifically, we computed the binary variables of ‘remitted‐TRS’ status and ‘non‐remitted (symptomatic) TRS’ status by taking the control group status as the reference. We repeated the GEE modeling using the same set of independent variables, cluster variable, covariates, and outcome variable as in Models 1–3, but only included our TRS patients and control participants as the sample. In supplementary analyses, we directly compared the ‘non‐remitted (symptomatic) TRS’ group with the ‘remitted TRS’ group using GEE models, with the ‘remitted‐TRS’ group as the reference (see Supplementary Tables S1–S3). We also examined the correlations of negative symptoms with SID performances in TRS patients and remitted schizophrenia patients separately (see Supplementary Tables S4–S6). Results Table 1 shows the sample characteristics. Among the TRS groups, 12, 28, and 16 participants had ‘early‐emerged TRS’ (<1 year after schizophrenia onset), ‘medium‐emerged TRS’ (2–5 years), and ‘late‐emerged TRS’ (>5 years) respectively 1 ; whilst the remaining four participants' onset of TRS could not be clearly ascertained. The TRS, remitted schizophrenia, and control groups were matched in age, sex ratio, and education (ps >0.05). However, TRS (P = 0.001) and remitted schizophrenia (P = 0.031) participants had lower non‐verbal IQ than controls. All participants scored <7 in the 17‐item HAM‐D, indicating absence of depression. 32 As expected, TRS participants had longer illness duration (mean = 96.3 months, SD = 49.95) than remitted schizophrenia participants (mean = 62.03 months, SD = 55.32) (t = 3.561, P = 0.001). All schizophrenia patients were taking antipsychotic medications, and all except 5 TRS participants were receiving clozapine. The antipsychotic dosages (DDD) between the TRS and remitted schizophrenia groups were comparable (t = 0.218, P = 0.828). Moreover, TRS participants had significantly higher anticholinergic medication burden than remitted schizophrenia participants (t = 4.979, P = 0.001). On the other hand, TRS and remitted schizophrenia participants also showed comparable drug compliance, as reflected in the DAI (t = 0.997, P = 0.321) and the MARS (t = 0.604, P = 0.547). Besides, the two schizophrenia groups both exhibited minimal levels of EPSE, as reflected in the ESRS (t = −0.509, P = 0.612) and the AIMS (t = 1.295, P = 0.198). Table 1Demographic and clinical information of the participantsTRS patients (N = 60)Remitted SCZ (N = 60)Controls (N = 60)F[2177] /t(118)/χ 2 P Post‐hoc (Sidak) pairwise comparisonMeanSDMeanSDMeanSDTRS vs. Remitted SCZ (Cohen's d)TRS vs. Controls (Cohen's d)Remitted SCZ vs. Controls (Cohen's d)Age (years)31.136.9331.538.1531.808.060.1130.893Sex (male vs. female)34 vs. 2622 vs. 3830 vs. 304.988* 0.083Education (years)13.033.1113.002.5513.522.890.6130.543TONI nonverbal IQ94.209.2296.388.89100.8810.397.684 0.001 0.509 (−0.241) 0.001 (−0.680) 0.031 (−0.465)Duration of treatment (months)77.4543.2042.2540.644.597 <0.001 Illness duration months96.3049.9562.0355.323.561 0.001 DDD (mg)15.4411.2315.067.380.2180.828Anticholinergic burden3.701.232.421.584.979 0.001 Number of psych admissions1.581.240.900.933.412 <0.001 PANSS positive9.573.857.601.433.709 <0.001 PANSS negative14.956.5711.284.793.492 0.001 PANSS general19.203.9518.032.211.998 0.048 CAINS MAP16.708.6410.486.144.545 <0.001 CAINS EXP7.354.243.123.545.934 <0.001 DAI total score13.202.4412.772.320.9970.321MARS composite score7.371.337.231.390.6040.547HAMD total0.651.420.571.140.180.832.7710.065ESRS total1.322.441.532.22−0.5090.612AIMS total0.180.600.070.361.2950.198SOFAS60.8515.4866.4814.0590.282.43699.192 <0.001 0.035 (−0.381) <0.001 (−2.657) <0.001 (−2.360)AIMS, the Abnormal Involuntary Movement Scale; CAINS, the Clinical Assessment Interview for Negative Symptoms; DAI, the Drug Attitude Inventory; DDD, daily defined dose; ESRS, the Extrapyramidal Side‐effect Rating Scale; HAM‐D, the 17‐item Hamilton Rating Scale for Depression; MARS, the Medication Adherence Rating Scale; PANSS, the Positive and Negative Symptom Scales; SCZ, schizophrenia; SOFAS, the Social and Occupational Functioning Assessment Scale; TRS, treatment‐resistant schizophrenia.* χ 2 test; significant P‐values are bold. Demographic and clinical information of the participants AIMS, the Abnormal Involuntary Movement Scale; CAINS, the Clinical Assessment Interview for Negative Symptoms; DAI, the Drug Attitude Inventory; DDD, daily defined dose; ESRS, the Extrapyramidal Side‐effect Rating Scale; HAM‐D, the 17‐item Hamilton Rating Scale for Depression; MARS, the Medication Adherence Rating Scale; PANSS, the Positive and Negative Symptom Scales; SCZ, schizophrenia; SOFAS, the Social and Occupational Functioning Assessment Scale; TRS, treatment‐resistant schizophrenia. χ 2 test; significant P‐values are bold. Similar to our previous results, 7 TRS participants showed higher levels of negative symptoms than remitted schizophrenia participants in CAINS‐MAP (t = 4.545, P < 0.001) and CAINS‐EXP (t = 5.934, P < 0.001). TRS participants also showed higher levels of PANSS‐positive (t = 3.709, P < 0.001) and general (t = 3.492, P = 0.001) symptoms than remitted schizophrenia patients. Among our TRS participants, 37 of them met the Andreasen's symptom remission criteria, 23 and therefore were deemed clozapine‐responsive. As expected, the three groups differed in SOFAS (F[2177] = 99.192, P < 0.001), with TRS participants having significantly poorer social functioning than remitted schizophrenia participants (P = 0.035), whilst remitted schizophrenia participants having poorer social functioning than controls (P < 0.001). Social motivation Table 2 shows the performance variables of SID task. The GEE model results for the entire sample are shown in Tables 3, 4, 5. In the GEE models, neutral cue was the reference to positive cue (social reward) and negative cue (social punishment); the control group served as the reference to the TRS group status and remitted schizophrenia group status. In Model 1a, we found that application of negative cues promoted SID accuracy (β = 0.091, P = 0.003), and male sex was associated with lower SID accuracy (β = −0.164, P = 0.005). As trial number increased, the SID accuracy improved (β = 0.023, P < 0.001), suggesting practice effects. However, the TRS status, remitted schizophrenia status, and interactions between group status and cue valence (see Model 1b) all failed to have significant impact on SID accuracy. In Models 2a and 2b, we found that the interaction between TRS and social punishment had resulted in lower anticipatory pleasure (β = −0.028, P = 0.013). Moreover, the interaction between TRS and social reward showed significant impacts to reduce anticipatory pleasure (β = −0.032, P = 0.008). In control participants, the anticipated avoidance of social punishment (negative cue) and the anticipated social reward (positive cue) both resulted in an increase of anticipatory pleasure. The negative β‐values indicated that such phenomenon was diminished in TRS participants. In Model 3a, we found that the application of negative cues reduced consummatory pleasure (β = −0.014, P = 0.004), but positive cues promoted consummatory pleasure (β = 0.080, P < 0.001). As trial number increased, the consummatory pleasure increased (β = 0.001, P < 0.001). In Model 3b, the interaction between TRS and positive cue had significant impacts on reducing consummatory pleasure (β = −0.094, P < 0.001). In control participants, the successful avoidance of social punishment (negative cues) and the receipt of social reward (positive cues) both resulted in an increase of consummatory pleasure. The negative β‐value indicated that such phenomenon was diminished in TRS participants. Table 2The mean and SD of performance variables of the Social Incentive Delay (SID) task across participant groupsTRS patients (N = 60)Remitted SCZ patients (N = 60)Controls (N = 60)MeanSDMeanSDMeanSDAccuracyPositive cue0.5960.1200.6050.1160.6320.066Neutral cue0.6020.1320.6040.1310.6380.065Negative cue0.5940.1140.6110.1100.6450.056Reaction time (ms)Positive cue133.83841.933141.02745.402133.29122.990Neutral cue137.40341.950143.32845.492135.93227.308Negative cue135.68340.227143.25541.740139.21827.838Anticipatory pleasurePositive cue5.2881.2085.1461.0445.6000.880Neutral cue5.3481.2165.0941.0425.4780.844Negative cue5.2401.2135.1231.0085.5280.922Consummatory pleasurePositive cue5.5671.1675.7881.1056.0690.842Neutral cue5.4031.2075.2480.9885.2750.762Negative cue5.2431.2575.1951.0345.1280.839SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. The mean and SD of performance variables of the Social Incentive Delay (SID) task across participant groups SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Table 3Generalized Estimation Equation (GEE) model results for SID hit‐or‐notbeta‐coefficientSE P‐valueModel 1a (all participants)Subject8.86E‐058.52E‐050.299Negative cue9.08E‐023.04E‐02 0.003 Positive Cue−2.68E‐022.93E‐020.360Remitted SCZ−1.37E‐011.64E‐010.403TRS−2.85E‐012.08E‐010.170Sex (male)−1.64E‐015.78E‐02 0.005 Age9.57E‐043.59E‐030.606Trial number2.26E‐021.97E‐03 <0.001 TONI nonverbal IQ4.53E‐032.62E‐030.083Anticholinergic burden4.64E‐023.93E‐020.238DDD (mg)2.39E‐034.64E‐030.606Model 1b (all participants)Subject8.88E‐058.52E‐050.297Negative cue1.58E‐015.41E‐02 0.004 Positive Cue−3.31E‐025.02E‐020.509Remitted SCZ−1.25E‐011.73E‐010.472TRS−2.41E‐012.15E‐010.263Sex (male)−1.64E‐015.78E‐02 0.005 Age9.48E‐043.60E‐030.792Trial number2.26E‐021.98E‐03 <0.001 TONI nonverbal IQ4.52E‐032.62E‐030.084Anticholinergic burden4.65E‐023.93E‐020.237DDD (mg)2.38E‐034.64E‐030.607Remitted SCZ × Negative Cue−6.33E‐027.29E‐020.386Remitted SCZ × Positive Cue2.57E‐027.15E‐020.719TRS × Negative cue−1.27E‐017.44E‐020.089TRS × Positive Cue7.93E‐037.12E‐020.911Significant P‐values are bold.DDD, daily defined dose of antipsychotics; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for SID hit‐or‐not Significant P‐values are bold. DDD, daily defined dose of antipsychotics; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Table 4Generalized Estimation Equation (GEE) model results for anticipatory pleasurebeta‐coefficientSE P‐valueModel 2a (all participants)Subject2.90E‐053.41E‐050.396Negative cue−1.58E‐034.05E‐030.697Positive Cue7.01E‐035.42E‐030.196Remitted SCZ−3.78E‐027.60E‐020.619TRS1.17E‐026.98E‐020.867Sex (male)−5.09E‐033.18E‐020.873Age8.05E‐051.52E‐030.958Trial number2.16E‐042.97E‐040.468TONI nonverbal IQ−6.96E‐041.16E‐030.547Anticholinergic burden−1.57E‐021.27E‐020.214DDD (mg)2.83E‐032.42E‐030.242Model 2b (all participants)Subject2.96E‐053.39E‐050.383Negative cue9.99E‐037.89E‐030.205Positive Cue2.33E‐027.89E‐03 0.025 Remitted SCZ−2.91E‐027.68E‐020.705TRS3.31E‐027.15E‐020.644Sex (male)−6.74E‐033.17E‐020.832Age−2.38E‐051.52E‐030.988Trial number2.17E‐042.97E‐040.466TONI nonverbal IQ−7.03E‐041.15E‐030.541Anticholinergic burden−1.59E‐021.27E‐020.211DDD (mg)2.83E‐032.40E‐030.238Remitted SCZ × Negative Cue−5.41E‐031.03E‐020.601Remitted SCZ × Positive Cue−1.51E‐021.51E‐020.319TRS × Negative cue−2.76E‐021.11E‐02 0.013 TRS × Positive Cue−3.16E‐021.19E‐02 0.008 Significant P‐values are bold.DDD, daily‐defined dose of antipsychotics; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for anticipatory pleasure Significant P‐values are bold. DDD, daily‐defined dose of antipsychotics; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Table 5Generalized Estimation Equation (GEE) model results for consummatory pleasurebeta‐coefficientSE P‐valueModel 3a (all participants)Subject5.11E‐053.32E‐050.124Negative cue−1.42E‐024.91E‐03 0.004 Positive Cue8.00E‐021.00E‐02 <0.001 Remitted SCZ2.01E‐036.62E‐020.976TRS−2.05E‐026.29E‐020.744Sex (male)1.35E‐032.76E‐020.961Age−2.19E‐041.36E‐030.872Trial number1.34E‐033.13E‐04 <0.001 TONI nonverbal IQ−5.43E‐041.28E‐030.672Anticholinergic burden4.70E‐031.14E‐020.68DDD (mg)1.89E‐032.05E‐030.358Model 3b (all participants)Subject4.98E‐053.34E‐050.136Negative cue−1.20E‐029.15E‐030.191Positive Cue1.24E‐012.04E‐02 <0.001 Remitted SCZ1.71E‐026.82E‐020.802TRS3.47E‐026.40E‐020.587Sex (male)3.24E‐032.79E‐020.908Age−4.54E‐041.37E‐030.741Trial number1.34E‐033.13E‐04 <0.001 TONI nonverbal IQ−5.76E‐041.30E‐030.657Anticholinergic burden4.51E‐031.13E‐020.691DDD (mg)2.03E‐032.08E‐030.329Remitted SCZ × Negative Cue5.20E‐031.28E‐020.684Remitted SCZ × Positive Cue−3.26E‐022.60E‐020.211TRS × Negative cue−1.13E‐021.22E‐020.356TRS × Positive Cue−9.41E‐022.42E‐02 <0.001 Significant P‐values are bold.DDD, daily‐defined dose of antipsychotics; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for consummatory pleasure Significant P‐values are bold. DDD, daily‐defined dose of antipsychotics; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. The GEE model results for the clinical sample evaluated the impacts of negative and other symptoms on social motivation processing in TRS patients (see Tables 6, 7, 8). In Model 4a, we found higher CAINS‐EXP was associated with lower SID accuracy (β = −0.037, P = 0.006) in TRS and remitted schizophrenia participants, suggesting that negative symptoms could affect social motivation processing. As expected, interactions between TRS group status and cue valence (see Model 4b) did not show any significant impact on SID accuracy, after controlling for the CAINS‐MAP and CAINS‐EXP. In Models 5a and 6a, the CAINS‐MAP and CAINS‐EXP did not show any significant impacts on TRS and remitted schizophrenia participants' anticipatory pleasure and consummatory pleasure in the SID task (ps > 0.05). As shown in Model 5b, the interaction between TRS and social punishment remained significant, even after controlling for the PANSS‐positive symptoms, PANSS‐general symptoms, and CAINS negative symptoms (β = −0.022, P = 0.026). In remitted schizophrenia participants (the reference group), the anticipated avoidance of social punishment (negative cue) resulted in reduction of anticipatory pleasure. The negative β‐values indicated that such phenomenon was diminished in TRS participants, relative to remitted schizophrenia participants. Lastly, Model 6b showed that the interaction between TRS and social reward remained significant, even after controlling for the PANSS‐positive symptoms, PANSS‐general symptoms, and CAINS negative symptoms (β = −0.062, P = 0.003). In remitted schizophrenia participants (the reference group), the receipt of social reward (positive cue) resulted in higher consummatory pleasure. The negative β‐values indicated that such phenomenon was diminished in TRS participants, relative to remitted schizophrenia participants. Table 6Generalized Estimation Equation (GEE) model results for SID hit‐or‐not (with clinical symptoms as covariates)beta‐coefficientSE P‐valueModel 4a (TRS patients + remitted SCZ patients)Subject3.46E‐051.01E‐040.731Negative cue6.50E‐023.64E‐020.074Positive Cue−2.46E‐023.62E‐020.498TRS−7.13E‐028.73E‐020.414Sex (male)−2.75E‐017.53E‐02 <0.001 Age3.36E‐034.80E‐030.484Trial number2.32E‐022.45E‐03 <0.001 TONI nonverbal IQ4.68E‐033.41E‐030.170Anticholinergic burden4.89E‐023.40E‐020.151DDD (mg)4.54E‐034.45E‐030.308PANSS‐positive−1.57E‐031.31E‐020.904PANSS‐general−2.27E‐021.75E‐020.193CAINS‐MAP1.40E‐027.69E‐030.070CAINS‐EXP−3.67E‐021.33E‐02 0.006 Model 4b (TRS patients + remitted SCZ patients)Subject3.46E‐051.01E‐040.731Negative cue9.69E‐025.04E‐020.055Positive Cue−7.57E‐035.14E‐020.883TRS−3.91E‐021.00E‐010.697Sex (male)−2.75E‐017.53E‐02 <0.001 Age3.35E‐034.81E‐030.485Trial number2.32E‐022.45e‐03 <0.001 TONI nonverbal IQ4.67E‐033.41E‐030.170Anticholinergic burden4.90E‐023.40E‐020.150DDD (mg)4.54E‐034.45E‐030.308PANSS‐positive−1.57E‐031.31E‐020.905PANSS‐general−2.27E‐021.75E‐020.193CAINS‐MAP1.40E‐027.70E‐030.070CAINS‐EXP−3.66E‐021.33E‐02 0.006 TRS × Negative cue−6.39E‐027.22E‐020.376TRS × Positive Cue−3.42E‐027.24E‐020.637Significant P‐values are bold.CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for SID hit‐or‐not (with clinical symptoms as covariates) Significant P‐values are bold. CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Table 7Generalized Estimation Equation (GEE) model results for anticipatory pleasure (with clinical symptoms as covariates)beta‐coefficientSE P‐valueModel 5a (TRS patients + remitted SCZ patients)Subject7.78E‐054.73E‐050.100Negative cue−6.69E‐034.76E‐030.160Positive Cue−1.71E‐046.19E‐030.978TRS3.95E‐025.56E‐020.477Sex (male)1.12E‐024.11E‐020.785Age−1.45E‐032.13E‐030.498Trial number2.00E‐043.66E‐040.584TONI nonverbal IQ−5.24E‐041.47E‐030.721Anticholinergic burden−1.35E‐021.37E‐020.325DDD (mg)2.28E‐032.45E‐030.351PANSS‐positive1.10E‐027.25E‐030.127PANSS‐general−5.69E‐036.53E‐030.384CAINS‐MAP−2.37E‐033.30E‐030.473CAINS‐EXP−7.33E‐066.59E‐030.999Model 5b (TRS patients + remitted SCZ patients)Subject7.79E‐054.73E‐050.099Negative cue4.54E‐036.25E‐030.468Positive Cue8.22E‐031.10E‐020.453TRS5.23E‐025.65E‐020.354Sex (male)−1.23E‐024.09E‐020.763Age−1.50E‐032.12E‐030.481Trial number2.01E‐043.66E‐040.583TONI nonverbal IQ−5.21E‐041.46E‐030.721Anticholinergic burden−1.38E‐021.37E‐020.312DDD (mg)2.28E‐032.44E‐030.350PANSS‐positive1.12E‐027.20E‐030.122PANSS‐general−5.59E‐036.51E‐030.391CAINS‐MAP−2.30E‐033.29E‐030.485CAINS‐EXP−3.39E‐046.59E‐030.959TRS × Negative cue−2.22E‐029.99E‐03 0.026 TRS × Positive Cue−1.65E‐021.24E‐020.182Significant P‐values are bold.CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for anticipatory pleasure (with clinical symptoms as covariates) Significant P‐values are bold. CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Table 8Generalized Estimation Equation (GEE) model results for consummatory pleasure (with clinical symptoms as covariates)beta‐coefficientSE P‐valueModel 6a (TRS patients + remitted SCZ patients)Subject1.16E‐044.64E‐05 0.013 Negative cue−1.48E‐025.87E‐03 0.012 Positive Cue6.10E‐021.08E‐02 <0.001 TRS−1.45E‐024.97E‐020.770Sex (male)−2.26E‐023.42E‐020.508Age−1.40E‐031.76E‐030.427Trial number1.43E‐034.17E‐04 <0.001 TONI nonverbal IQ−4.91E‐051.72E‐030.977Anticholinergic burden8.34E‐031.20E‐020.486DDD (mg)1.87E‐032.00E‐030.351PANSS‐positive8.93E‐036.73E‐030.185PANSS‐general−7.54E‐037.66E‐030.325CAINS‐MAP−4.78E‐033.74E‐030.202CAINS‐EXP−2.92E‐047.02E‐030.967Model 6b (TRS patients + remitted SCZ patients)Subject1.18E‐044.64E‐05 0.011 Negative cue−6.52E‐038.92E‐030.465Positive Cue9.17E‐021.63E‐02 <0.001 TRS2.41E‐025.12E‐020.638Sex (male)−2.47E‐023.45E‐020.474Age−1.67E‐031.77E‐030.346Trial number1.44E‐034.16E‐04 0.001 TONI nonverbal IQ−3.90E‐051.74E‐030.982Anticholinergic burden7.81E‐031.19E‐020.512DDD (mg)1.90E‐032.01E‐030.345PANSS‐positive9.53E‐036.74E‐030.157PANSS‐general−7.52E‐037.65E‐030.326CAINS‐MAP−5.20E‐033.76E‐030.166CAINS‐EXP1.37E‐057.07E‐030.998TRS × Negative cue−1.64E‐021.20E‐020.171TRS × Positive Cue−6.18E‐022.08E‐02 0.003 Significant P‐values are bold.CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for consummatory pleasure (with clinical symptoms as covariates) Significant P‐values are bold. CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; SCZ, schizophrenia; TRS, treatment‐resistant schizophrenia. The GEE model results with the differentiation between the remitted‐TRS and non‐remitted (symptomatic) TRS status further clarified the impacts of symptom remission on social motivation processing in TRS patients (see Tables 9, 10, 11). In Model 7b, the remitted‐TRS status, non‐remitted (symptomatic) TRS status, and interactions between the TRS group status and cue valence all failed to have a significant impact on SID accuracy, consistent with our earlier GEE model results. In Model 8b, we found that the interaction between remitted‐TRS status and social punishment still showed significant impacts to reduce anticipatory pleasure (β = −0.042, P = 0.013), and the interaction between remitted‐TRS status and social reward also still resulted in lower anticipatory pleasure (β = −0.039, P = 0.019), albeit the limited sample size of the remitted‐TRS group (n = 37). In control participants, the anticipated avoidance of social punishment (negative cue) and the anticipated social reward (positive cue) both resulted in an increase of anticipatory pleasure. The negative β‐values indicated that such phenomenon was diminished in remitted‐TRS participants. Regarding the impacts of non‐remitted (symptomatic) TRS status on social motivation processing, the interaction between non‐remitted‐TRS status and social reward had significantly reduced anticipatory pleasure (β = −0.028, P = 0.016). However, the interaction between non‐remitted (symptomatic) TRS status and social reward failed to have significant impacts on anticipatory pleasure (P > 0.05), likely related to the limited sample size of the non‐remitted (symptomatic) TRS group (n = 23). In Model 9b, we found that both the interaction between remitted‐TRS status and positive cue (β = −0.110, P = 0.002), and the interaction between non‐remitted (symptomatic) TRS status and positive cue (β = −0.085, P < 0.001) remained having significant impacts on reducing consummatory pleasure, albeit the limited sample size. In control participants, the successful avoidance of social punishment (negative cues) and the receipt of social reward (positive cues) both resulted in an increase of pleasure. The negative β‐value indicated that such phenomenon was diminished in both remitted‐TRS and non‐remitted (symptomatic) TRS participants. Table 9Generalized Estimation Equation (GEE) model results for SID hit‐or‐not (with differentiation between remitted‐TRS and non‐remitted TRS status)beta‐coefficientSE P‐valueModel 7a (Remitted TRS patients + non‐remitted TRS patients + controls)Subject5.00E‐051.10E‐040.648Negative cue8.50E‐023.79E‐02 0.025 Positive Cue−3.68E‐023.56E‐020.301Remitted TRS−3.41E‐014.02E‐010.397Non‐remitted TRS−2.28E‐014.27E‐010.594Sex (male)−1.62E‐017.71E‐02 0.035 Age−1.38E‐034.38E‐030.753Trial number2.14E‐022.39E‐03 <0.001 TONI nonverbal IQ5.20E‐032.96E‐030.079Anticholinergic burden2.68E‐029.74E‐020.783DDD (mg)3.51E‐036.54E‐030.591Model 7b (Remitted TRS patients + non‐remitted TRS patients + controls)Subject5.05E‐051.10E‐040.645Negative cue1.54E‐015.42E‐02 0.004 Positive cue−3.27E‐025.01E‐020.514Remitted TRS−2.66E‐014.16E‐010.522Non‐remitted TRS−2.01E‐014.28E‐010.638Sex (male)−1.63E‐017.71E‐02 0.035 Age−1.38E‐034.39E‐030.754Trial number2.14E‐022.40E‐03 <0.001 TONI nonverbal IQ5.19E‐032.96E‐030.079Anticholinergic burden2.67E‐029.75E‐020.784DDD (mg)3.51E‐036.53E‐030.592Remitted TRS × Negative Cue−1.15E‐011.04E‐010.270Remitted TRS × Positive Cue−1.07E‐019.87E‐020.277Non‐remitted TRS × Negative cue−1.32E‐018.16E‐020.105Non‐remitted TRS × Positive Cue5.36E‐027.85E‐020.495Significant P‐values are bold.CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for SID hit‐or‐not (with differentiation between remitted‐TRS and non‐remitted TRS status) Significant P‐values are bold. CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; TRS, treatment‐resistant schizophrenia. Table 10Generalized Estimation Equation (GEE) model results for anticipatory pleasure (with differentiation between remitted‐TRS and non‐remitted TRS status)beta‐coefficientSE P‐valueModel 8a (Remitted TRS patients + non‐remitted TRS patients + controls)Subject4.96E‐055.00E‐050.322Negative cue−3.51E‐035.30E‐030.507Positive Cue6.27E‐035.97E‐030.293Remitted TRS1.72E‐011.23E‐010.161Non‐remitted TRS1.48E‐019.89E‐020.135Sex (male)1.59E‐023.99E‐020.692Age2.15E‐042.14E‐030.920Trial number6.45E‐043.63E‐040.075TONI nonverbal IQ−1.46E‐031.50E‐030.331Anticholinergic burden−5.37E‐022.33E‐02 0.021 DDD (mg)3.80E‐033.34E‐030.255Model 8b (Remitted TRS patients + non‐remitted TRS patients + controls)Subject5.00E+004.95E‐050.316Negative cue1.13E‐027.81E‐030.147Positive Cue2.33E‐021.05E‐02 0.026 Remitted TRS1.96E‐011.27E‐010.123Non‐remitted TRS1.62E‐011.02E‐010.111Sex (male)1.29E‐023.98E‐020.746Age3.71E‐062.14E‐030.999Trial number6.47E‐043.63E‐040.075TONI nonverbal IQ−1.49E‐031.49E‐030.320Anticholinergic burden−5.35E‐022.40E‐02 0.026 DDD (mg)3.85E‐033.32E‐030.246Remitted TRS × Negative Cue−4.20E‐021.68E‐02 0.013 Remitted TRS × Positive Cue−3.88E‐021.66E‐02 0.019 Non‐remitted TRS × Negative cue−1.87E‐021.15E‐020.104Non‐remitted TRS × Positive Cue−2.75E‐021.14E‐02 0.016 Significant P‐values are bold.CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for anticipatory pleasure (with differentiation between remitted‐TRS and non‐remitted TRS status) Significant P‐values are bold. CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; TRS, treatment‐resistant schizophrenia. Table 11Generalized Estimation Equation (GEE) model results for consummatory pleasure (with differentiation between remitted‐TRS and non‐remitted TRS status)beta‐coefficientSE P‐valueModel 9a (Remitted TRS patients + non‐remitted TRS patients + controls)Subject5.64E‐054.56E‐050.216Negative cue−1.72E‐025.83E‐03 0.003 Positive Cue7.37E‐021.26E‐02 <0.001 Remitted TRS2.46E‐031.03E‐010.981Non‐remitted TRS3.39E‐029.09E‐020.709Sex (male)3.13E‐023.39E‐020.357Age−7.16E‐041.86E‐030.700Trial number1.66E‐033.72E‐04 <0.001 TONI nonverbal IQ−1.14E‐031.53E‐030.457Anticholinergic burden2.80E‐042.40E‐02 0.991 DDD (mg)4.95E‐042.96E‐030.867Model 9b (Remitted TRS patients + non‐remitted TRS patients + controls)Subject5.42E‐054.59E‐050.237Negative cue−1.10E‐029.16E‐030.230Positive Cue1.24E‐012.04E‐02 <0.001 Remitted TRS6.53E‐021.05E‐010.533Non‐remitted TRS7.69E‐029.20E‐020.403Sex (male)3.49E‐023.47E‐020.315Age−1.10E‐031.88E‐030.560Trial number1.67E‐033.71E‐04 <0.001 TONI nonverbal IQ−1.19E‐031.54E‐030.441Anticholinergic burden1.36E‐042.40E‐020.996DDD (mg)7.57E‐043.02E‐030.802Remitted TRS × Negative Cue−2.67E‐021.71E‐020.118Remitted TRS × Positive Cue−1.10E‐013.50E‐02 0.002 Non‐remitted TRS × Negative cue−2.12E‐031.30E‐020.871Non‐remitted TRS × Positive Cue−8.52E‐022.36E‐02 <0.001 Significant P‐values are bold.CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; TRS, treatment‐resistant schizophrenia. Generalized Estimation Equation (GEE) model results for consummatory pleasure (with differentiation between remitted‐TRS and non‐remitted TRS status) Significant P‐values are bold. CAINS, the Clinical Assessment Interview for Negative Symptoms; DDD, daily‐defined dose; PANSS, the Positive and Negative Symptom Scales; TRS, treatment‐resistant schizophrenia. Discussion Although impaired social motivation has not been regarded as one of the core features for TRS, 37 some historical constructs related to treatment non‐response, such as hebephrenic schizophrenia and deficit syndrome of schizophrenia, 38 are characterized by negative symptoms, including social anhedonia. The latest TRRIP criteria 1 advocates the characterization of TRS patients based on positive, negative and cognitive symptoms, to identify possible intervention targets for improving patients' social functioning and clinical outcomes. This study is among the first few research to characterize the nature and extent of social motivation processing in TRS patients, using an experimental paradigm and GEE models. Although social reward and social punishment in everyday life can be far more complicated than the cartoon images used in the SID task, our behavioral paradigm has tapped into participants' motivation processing, in view of the observed significant association of negative cues with the outcome of ‘hit‐or‐not’, and the significant association of social reward/punishment with consummatory pleasure. Moreover, other social cognitive paradigms 39 , 40 have also used cartoon‐based images (e.g. emoji) to present social‐salient information. Therefore, the cartoon‐based images presented in the outcome phase of the SID task are believed to contain social‐relevant information, and could motivate participants' task performance. Our results suggested that TRS, remitted schizophrenia, and interaction between TRS/remitted schizophrenia and valence of cue did not have significant impacts on SID accuracy. In motivation processing, 3 , 15 , 41 important components include cost–benefit computation, anticipatory and consummatory pleasure, action‐plan formation, and execution of action. Among the three SID performance variables used in our study (i.e. ‘hit‐or‐not’, anticipatory pleasure, consummatory pleasure), the SID accuracy reflected participants' executed action within a short response window, and might not be a sensitive index to discriminate TRS patients and remitted schizophrenia patients from healthy controls. Other behavioral paradigms, such as effort‐based decision‐making task with social scenarios, 15 may better capture effort‐expenditure with high sensitivity. On the other hand, we found that the anticipated avoidance of social punishment and anticipated social reward could bring significantly less increment of anticipatory pleasure to TRS patients than healthy controls, as evidenced by the significant interaction between TRS group status and valence of cues in the GEE model for anticipatory pleasure. Similarly, the receipt of social reward could bring significantly less increment of consummatory pleasure to TRS patients than healthy controls. Together, these findings indicate that TRS is associated with altered anticipatory pleasure and consummatory pleasure in social motivation processing, relative to healthy people. Our findings also extended the previous research on non‐social motivation deficits in TRS patients 41 to social motivation processing. Consistent with our hypothesis, negative symptoms showed a significantly negative impact on social motivation processing in TRS patients and remitted schizophrenia patients, as evidenced by the significant effects of the CAIN‐EXP on SID ‘hit‐or‐not’ outcome in the GEE models. Nevertheless, the CAINS‐MAP and CAIN‐EXP did not appear to influence other SID performance variables, including anticipatory pleasure and consummatory pleasure in TRS and remitted schizophrenia patients. These unexpected results may be related to the fact that the majority (n = 97) of our clinical sample had fulfilled the Andreasen's symptom remission criteria, 23 and exhibited minimal levels of negative symptoms. Given that healthy controls were not measured with the CAINS, the GEE models which controlled for the impacts of negative symptoms on SID performance only included clinical participants. After controlling for negative (and other) symptoms, TRS patients still anticipated significantly less pleasure for the anticipated avoidance of social punishment and experienced less consummatory pleasure for the receipt of social reward, relative to remitted schizophrenia patients. Moreover, after differentiating the remitted‐TRS status from the non‐remitted (symptomatic) TRS status, the GEE models showed that remitted‐TRS patients still exhibited altered anticipatory pleasure and consummatory pleasure in social motivation processing, relative to healthy people, as evidenced by the significant interaction between remitted‐TRS group status and valence of cues in the GEE models for anticipatory and consumatory pleasure. Together, such findings support our hypothesis that TRS patients have altered social motivation processing, regardless of remission status and even after controlling for the effects of negative and other symptoms. Our preliminary findings of altered social motivation processing in TRS patients may have clinical implications. Around 23% of first‐episode schizophrenia patients would eventually develop treatment‐resistance, 42 but the effectiveness of current evidence‐based treatments for TRS remains unsatisfactory. 43 Many TRS patients who responded to clozapine continued to have poor functional outcomes. 44 Our findings lend support to the clinical potentials of intervening social motivation processing to benefit TRS patients' social functioning. Although cognitive remediation has been reported to improve functional outcomes of TRS patients, 45 such intervention does not target at social motivation. Motivational interviewing coupled with cognitive behavioral therapy 46 and cognitive‐behavioral social‐skill training 47 have been proposed as potential interventions for improving social motivation in schizophrenia patients. 10 Future study should explore the feasibility of applying these strategies to TRS patients, and evaluate its impacts on negative symptoms and social functioning. Furthermore, our work may have theoretical implications. Interestingly, motivation has been recently reported to be related to treatment engagement and treatment response in schizophrenia patients. 48 It is plausible that impaired social motivation processing may contribute to the emergence of negative symptoms in schizophrenia patients and may undermine patients' treatment response to antipsychotics. A recent study examined whether early emergence of negative symptoms could predict future development of TRS, but the results were largely non‐significant. 49 However, such negative findings may be limited by the crude measure of negative symptoms using a first‐generation scale and the lack of experimental paradigms to tap into cognitive and motivation processing closely related to negative symptoms. Future research should further clarify whether social motivation would predict negative symptoms and non‐response to treatments. Although the ‘liking‐wanting anhedonia paradox’ suggested that schizophrenia patients tended to have intact ‘liking’ but impaired ‘wanting’, 50 recent meta‐analytic evidence indicated that schizophrenia patients might have diminished ‘social’ consummatory pleasure, relative to healthy people. 51 Our findings supported alterations of social motivation processing in TRS patients relative to remitted schizophrenia and controls, affecting anticipatory and consummatory pleasure. Such findings appeared to concur with the current notions. 50 , 51 On the other hand, remitted schizophrenia patients did not differ in social motivation processing than controls, contrary to prior research using other paradigms for social motivation processing (such as effort‐based decision‐making tasks). 14 , 15 The divergent findings might be related to the fact that all remitted schizophrenia participants had fulfilled the Andreasen's symptom remission criteria, 23 and exhibited minimal levels of symptoms. Prior research on social motivation processing in schizophrenia patients did not recruit samples in full clinical remission. 14 , 15 Another possible reason may be related to the limitations of the SID paradigm, 16 , 17 which only used cartoon‐based social reward and punishment. This study has several limitations. Although our TRS participants had shorter illness duration than many previous chronic TRS samples, 7 , 8 , 41 , 44 they still had a considerable length of illness duration, longer than the remitted schizophrenia participants. Moreover, social isolation had been found to affect social approach motivation, 52 but it was not measured in this study. Future study may recruit recently‐emerged TRS patients and measure participants' levels of loneliness to account for these confounds. Third, although the two clinical groups had comparable anticholinergic burden, antipsychotic‐induced parkinsonism side‐effects, and depressive symptoms, we did not control for antipsychotic‐induced sedation, which might lead to secondary negative symptoms affecting social motivation processing. Fourth, the SID task is not a sophisticated task to capture social motivation clearly, which might have resulted in the non‐significant impacts of TRS/remitted schizophrenia and the interaction between TRS/remitted schizophrenia and the valence of cue on the SID accuracy. More sophisticated behavioral paradigms, such as the effort‐based decision‐making task with social scenarios, 15 should be used in future research to tap into social motivation processing. Fifth, this study did not employ a non‐social behavioral task; therefore, we could not make a comparison between social and non‐social motivation processing in TRS patients. Lastly, this study was cross‐sectional, and we could not infer the contribution of social motivation deficits to negative symptoms. To conclude, using an experimental paradigm and the more sophisticated GEE models, our preliminary findings suggested that TRS patients exhibited impaired social motivation processing, affecting their ability to anticipate and experience pleasure to social reward/punishment. Future research should adopt a more sophisticated behavioral task to clarify whether social motivation deficits in schizophrenia may predict the risk for developing TRS and functional impairments. Author contributions Simon S.Y. Lui: Conceptualization, Formal analysis, Methodology, Supervision, Visualization, Writing – original draft, Writing –review and editing. Jasmine W.S. Chan: Conceptualization, Data curation, Investigation, Formal analysis, Methodology, Project administration, Writing – original draft, Writing –review and editing. Jason L.F. Chan: Data curation, Investigation, Writing – review and editing. Kimi H.Y. Lam: Data curation, Investigation, Writing – review and editing. Lok‐Yin Choi: Formal analysis, Visualization, Writing – review and editing. Raisie W.K. Wong: Data curation, Investigation, Writing – review and editing. Perry B.M. Leung: Validation, Writing – review and editing. Na Zhan: Writing – review and editing. Jason W.Y. Wong: Project administration, Resources. Mindi W.Y. Chiu: Project administration, Methodology, Resources. Jenny P.H. Lam: Project administration, Resources. Raymond C.K. Chan: Conceptualization, Methodology, Supervision, Writing – review and editing. Disclosure statement The authors declare no conflicts of interest with respect to the authorship or the publication of this article. Supporting information Table S1. Generalized Estimation Equation (GEE) model results for SID hit‐or‐not to compare remitted TRS with non‐remitted TRS patients. Table S2. Generalized Estimation Equation (GEE) model results for anticipatory pleasure to compare remitted TRS with non‐remitted TRS patients. Table S3. Generalized Estimation Equation (GEE) model results for consummatory pleasure to compare remitted TRS with non‐remitted TRS patients. Table S4. Spearman's correlations of change in mean scores of SID performance with the CAINS subscale scores in the clinical sample (n = 120). Table S5. Spearman's correlations of change in mean SID anticipatory pleasure ratings with the CAINS subscale scores in TRS sample (n = 60). Table S6. Spearman's correlations of change in mean SID anticipatory pleasure ratings with the CAINS subscale scores in remitted schizophrenia sample (n = 60). Acknowledgements This work is supported by HKU Seed Fund for PI Research – Basic Research (Project code: 2202100748) granted to Lui SSY; and the Philip K. H. Wong Foundation donated to Chan RCK. The funding agents have no role in the study design; collection, analysis, and interpretation of the data; writing of the manuscript; nor decision to submit the paper for publication. We thank our funding agents and all the participants of this study. Contributor Information Simon S.Y. Lui, Email: lsy570@hku.hk. Raymond C.K. Chan, Email: rckchan@hku.hk. Data availability statement The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. References Associated Data Supplementary Materials Table S1. Generalized Estimation Equation (GEE) model results for SID hit‐or‐not to compare remitted TRS with non‐remitted TRS patients. Table S2. Generalized Estimation Equation (GEE) model results for anticipatory pleasure to compare remitted TRS with non‐remitted TRS patients. Table S3. Generalized Estimation Equation (GEE) model results for consummatory pleasure to compare remitted TRS with non‐remitted TRS patients. Table S4. Spearman's correlations of change in mean scores of SID performance with the CAINS subscale scores in the clinical sample (n = 120). Table S5. Spearman's correlations of change in mean SID anticipatory pleasure ratings with the CAINS subscale scores in TRS sample (n = 60). Table S6. Spearman's correlations of change in mean SID anticipatory pleasure ratings with the CAINS subscale scores in remitted schizophrenia sample (n = 60). Data Availability Statement The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.