Can antipsychotics improve social cognition in patients with schizophrenia?

Social cognition is described as the higher mental processes that are engaged while people store, process, and use social information to make sense of themselves and others. Aspects of social cognition include emotion perception, social cue interpretation, attribution style, and theory of mind, all of which appear disordered in schizophrenia. Such social cognitive deficits are believed to be important predictors of functional outcome in schizophrenia, therefore they may represent a crucial treatment target. Few studies have evaluated the influence of antipsychotic treatment on these deficits. The purpose of this review is to examine the relationship between antipsychotic treatment and social cognition, whether antipsychotics improve social cognitive function, and if so to explore differential medication effects. Comprehensive searches of PsycINFO and MEDLINE/PUBMED were conducted to identify relevant published manuscripts. Fifteen relevant papers published in English were found, describing original studies. On the basis of this review, we have drawn the following conclusions: first, the results do not engender optimism for the possibility that antipsychotic drugs can specifically facilitate social recovery. Second, the actions of antipsychotics on social cognition are inconclusive, due to lack of standardization across research groups, leading to inconsistencies between study designs, methods used, and medication dosages. Third, large-scale longitudinal investigations are needed to explore the unclear relationships between social cognition, symptoms, and functional outcome. Other non-pharmacological treatments focusing on training patients in the social cognitive areas may hold more promise.

Introduction

Social cognition has been defined as the way we perceive, interpret, and understand social information [1] or as “the processes that allow a person to understand, act on, and benefit from the interpersonal world” [2]. A further overall definition was given by Adolphs [3] who described social cognition as “the ability to construct representations of the relation between oneself and others and to use those representations flexibly to guide social behaviour.”

Aspects of social cognition include emotion perception, social cue interpretation, attribution style, and theory of mind. Affect perception is the ability to infer emotional information, in other words what a person is feeling, presented either in visual or auditory form. Social cue perception refers to a person’s ability to ascertain social cues from behaviour provided in a social context, and refers to a person’s comprehension of social rules [4]. Attribution style, known as a personalizing bias, refers to an individual’s own perception and interpretation of facts and events [5]. The attribution of mental states, such as desires, intentions, and beliefs, to other people has been referred to as “theory of mind” (ToM) or “mentalising” [6, 7]. ToM involves both the ability to understand that others have mental states different from one’s own, and the capability to make correct inferences about the content of those mental states [5].

The neurophysiological and neurochemical underpinnings of social cognition in schizophrenia are a scientific domain that requires further exploration. Several neurotransmitters seem to play a considerable role in social cognitive processes, and their circuitries are deemed to be altered in schizophrenia. The hypothalamic peptides arginine vasopressin (AVP) and oxytocin (OXT) have been described as social hormones that may mediate social behaviour [8] including social motivation, approach behaviour [9], and ToM [10, 11]. Recently, abnormal oxytocinergic and dopaminergic signalling in the amygdalae has been proposed to explain dysfunction in the social cognitive domain in schizophrenia [12].

Serotonin is another neurotransmitter linked to social behaviour, including roles in cognition, mood, and aggression, alongside motivation, energy levels, and sleep [13, 14]. There is increasing interest in the correlation between negative symptoms of schizophrenia and abnormal neurotransmission at serotonin 5-HT2 receptors [15].

Dopamine appears as a key neurotransmitter in the aetiopathogenesis of schizophrenia described as crucially involved in the attribution process as well as emotional perception, giving not only meaning but also salience to the objects in our environment [16, 17]. Deregulation of the dopaminergic system leads to the production of dopamine regardless of incoming stimuli, which results in giving meaning to their meta-representations, thereby creating a misguided inner reality of actually meaningless objects. This maladaptive attribution system created during psychosis is very often implicated permanently in the patient’s experience, regardless of pharmacological blocking of excess dopamine [16].

A growing body of literature has shown consistently that schizophrenia patients compared with healthy controls present with social cognitive impairments that are relatively stable and persistent, suggesting that it is a trait-dependent rather than state-dependant aspect of the disorder [18-20]. These deficits have been widely described as modifying patients’ behaviour when interacting with other people (ToM deficits) [7, 20–22] and in recognizing emotions [23-25] and other social information cues [18, 26]. Therefore, social cognitive deficits are believed to be important predictors of functional outcome in schizophrenia [5, 27]. Such deficits represent an obvious substrate for treatment in schizophrenia.

Pharmacological treatment research on social cognition in schizophrenia has been relatively limited: recent data on the effects of second- and first-generation antipsychotics (SGA/FGA) on various domains of social cognition remain inconclusive.

This paper aims (1) to appraise current evidence on the impact of antipsychotics upon social cognitive functioning in schizophrenia, to find out if antipsychotics do really improve social cognitive functions; and if yes, (2) to explore differential medication effects on social cognition, if any.

A comprehensive search of the PsycINFO and MEDLINE/PUBMED databases for articles in English published till 31 December 2012 was conducted. Within the domain of social cognition, the following search terms were used: emotion/affect perception, emotion/affect recognition, attribution/attributional style, theory of mind/mentalising, social cognition, social competence, and social cue perception. Within the domain of psychopharmacology outcome, the following terms were used: conventional antipsychotics, atypical antipsychotics, atypicals, and clinical trials.

Search terms for schizophrenia included the following: psychosis, schizophrenia, and schizoaffective disorder.

Search Strategy

The following search keywords were used:

schizophrenia AND social cognition AND antipsychotics; 224 articles, 15 utilised

schizophrenia AND emotion perception AND antipsychotics; 63 papers, 6 utilised

schizophrenia AND facial affect AND antipsychotics; 29 papers, 4 utilised

schizophrenia AND theory of mind AND antipsychotics; 14 papers, 3 utilised

schizophrenia AND attribution AND antipsychotics; 20 articles, 0 utilised

schizophrenia AND attributional style AND antipsychotics; 3 articles, 1 utilised

schizophrenia AND social competence AND antipsychotics; 33 articles, 2 utilised

schizophrenia AND social cue perception AND antipsychotics; 2 papers, 1 utilised

Inclusion Criteria

The papers were utilised in the current review if they were written in English and had reported experimental studies of aspects of social cognition in schizophrenia treated with antipsychotic medication. Although 32 papers were identified there was much overlap between the results of individual searches: 15 papers in total were accrued from all eight searches and are reviewed here (see Table 1).

Table 1

Summary of the 15 papers reviewed

References	Study design	Subject groups with numbers	Antipsychotics used (CPZE daily ± SD)	Study duration	Evaluated social cognitive domain	Social cognitive tests	Main results	Criticism
Gaebel and Wölwer [23]	RCT	S = 23 first-onset	P = 376 mg	4 weeks	FAR	Ekman and Friesen	Comparable improvement in FAR in both patient groups	Small sample size
			H = 445 mg
		P = 13
		H = 10
		HC = 15
Lewis and Garver [34]	NCT	S = 18 chronic	H = 250–1,000 mg	2 weeks	FAR	Ekman and Friesen	No effect of haloperidol on FAR	Small sample size
		H = 18
								Non-randomised
		HC = 10
Kee et al. [32]	DBCT	S = 18 chronic	R = 300 mg	8 weeks	Emotion perception (FAR, emotion prosody)	FEIT	R > H	Small sample size
		R = 9	H = 750 mg			VEIT	No improvement of emotion perception in H group
		H = 9				VAPT
Williams et al. [36]	NCC	S = 28 chronic	R = 828 ± 493 mg	Assessed once	FAR	Colour photographs	HC > R > H	Small sample size
		R = 15	H = 688 ± 601 mg				Patients on risperidone performed worse than HC; however, better than H-treated group	Non-randomised Use of limited social cognitive measures
		H = 13
		HC = 28
Littrell et al. [44]	NCT	S = 52 chronic	FGAs = 540 ± 215 mg	12 month	Emotional perception, perception of social cue	IPT	Improvement in social cognitive domains in O-treated patients solely	Small sample size
		FGAs = 30 O = 22	Mean O dose = 15.3 mg/day	Assessed at baseline, 12 weeks, and end point (52 weeks)				Non-randomised
								No psychopathology rating
Herbener et al. [31]	NCT	S = 13 first-onset	R = 169 ± 100 mg	31.3 ± 8.3 days	FAR	PEAT	No effect of medications on FAR	Small sample size
		R = 9	H = 225 ± 35 mg			EMODIFF		Non-randomised
		H = 2	Z = 233 mg
		Z = 1	A = 400 mg
		A = 1
		HC = 13
Harvey et al. [29]	DBCT	S = 289 chronic	R = 100–400 mg	8 weeks	Emotion perception, social competence	PEAT	Both medications improved social competence but not emotion perception	Lack of non-medication control group
		R = 154	Q = 267–1,067 mg
		Q = 135
Savina and Beninger [45]	Cross-sectional	S = 84 chronic	Not provided	Assessed once	ToM	First-order belief and second-order belief tasks and faux pas tests	FGA and R groups performed worse than other groups on ToM tasks. O and C groups were comparable to HCs on ToM tasks	Small sample size Non-randomised
		FGAs = 23
		C = 18
		O = 20
		R = 3
		HC = 24
Sergi et al. [42]	DBCT	S = 73 chronic R = 32 O = 28 H = 13	R = 48.2 ± 7 mg O = 49.2 ± 6 mg H = 50.0 ± 5 mg	8 weeks	Emotion perception	FEIT VEIT HPNS IPT	No effect of medication on emotion perception	Lack of non-medication control group Modest group size Random assignment paths
Mizrahi et al. [21]	Cross-sectional component	Psychotic disorder = 71 SGA 88.6 % FGA 11.4 %	R = 3–4 mg	6 weeks (measurements every 2 weeks)	ToM	Hinting task	Improvement on both PANSS and ToM on antipsychotics	No placebo control group
	Longitudinal component	S = 17 first-onset (60 % neuroleptic naive; 40 % drug free)	O = 2.5–20 mg
			L = 35 mg
			C = 225–300 mg
Machado de Sousa and Hallak [43]	Cross-sectional	S = 15 C = 15 (resistant to treatment) chronic HC = 15	C = 470 ± 173 mg	Assessed once	FAR	ERT (based on pictures of facial affect (Ekman and Friesen))	Patients took more time to perform ERT Time-related deficits for recognition of fear and disgust in patients were found	Small sample size Non-randomised
Fakra et al. [46]	RCT	S = 25 chronic	R = 298 ± 116 mg	1 month	FAR	Feinberg test	R > H on FAR	Small sample size Lack of non-medication control group
		R = 11	H = 398 ± 155 mg
		H = 14
Penn et al. [33]	DBCT	S = 873 chronic	Identical-appearing capsules contained O = 7.5 mg; Q = 200 mg; R = 1.5 mg; Pph = 8 mg and Z = 40 mg The medication dose ranging from one to four capsules daily, based upon the study doctor’s judgement	18 months	Emotion perception	FEDT	Limited medication effect on emotion perception was found	Lack of non-medication control group
		O = 213
		Q = 54
		R = 183 Combination = 72l
		All others = 130
Roberts et al. [38]	DBCT	S = 223 chronic O = 117 Q = 106	O = 312 mg/day Q = 607 mg/day	6 months	Perception of social cues	Social cue recognition test	Improvement in both medication groups on 3 out of 4 social cognitive subscales	Lack of non-medication control group
Kucharska-Pietura et al. [35]	Cross-sectional	S = 84 chronic FGA = 28 (Pph = 14; H = 14) SGA = 56	FGA = 422 ± 219 SGA:  O = 341 ± 118  C = 519 ± 276	Assessed once	Emotion perception; ToM/empathy	FERT VERT Reading the Mind in the Eyes Test	There were no statistically significant differences on social cognitive performance between FGA and SGA treatment groups	Small sample size Non-randomized

Designs: DBCT double-blind clinical trial, HC healthy controls, NCC non-randomized case control study, NCT non-randomized clinical trial, RCT randomized clinical trial

Medications: A aripiprazole, C clozapine, CPZE chlorpromazine equivalent, FGAs first generation antipsychotics, H haloperidol, L loxapine, O olanzapine, P perazine, Pph perphenazine, Q quetiapine, R risperidone, SGAs second generation antipsychotics, Z ziprasidone

Subject groups: HC Healthy Controls, S schizophrenia group

Social cognitive measures: EMODIFF Penn Emotion Differentiation Test, ERT Emotion Recognition Test, FEIT Facial Emotion Identification Test, FEDT The Face Emotion Discrimination Task, HPNS Half-Profile of Nonverbal Sensitivity, FERT Facial Emotion Recognition Test, IPT Interpersonal Perception Task, PEAT Penn Emotional Acuity Test, Reading the Mind in the Eyes Test, VAPT Videotape Affect Perception Test, VEIT Voice Emotion Identification Test, VERT Voice Emotion Recognition Test

Social cognitive domains: FAR Facial Affect Recognition, ToM Theory of Mind

Psychiatric scale: PANSS Positive and Negative Syndrome Scale

The Place of Social Cognitive Deficit in Multifactorial Models of Schizophrenia: Symptom, or Neurocognitive Compromise?

The analysis of emotional behaviour in schizophrenia is fundamental to the notion of dementia praecox introduced by Kraepelin, and a question fielded by Bleuler on the basis of ‘Affektivität’: “What happened to feelings in dementia praecox?” This has constituted and still constitutes a scientific challenge [28]. While symptom-based approaches have understandably dominated most aspects of pharmacological intervention in schizophrenia research, a limited number of studies have investigated the effect of symptoms on social cognition [29-33].

To the best of the authors’ knowledge, the majority of studies fail to demonstrate a clear relationship between overall symptom severity scores and performance on social cognitive measures [29, 34–36].

However, there are suggestions that negative and disorganized symptoms may be related to social cognitive functioning [21, 23, 31, 37, 38].

Relationships among social cognitive constructs and negative symptoms are, however, not clear. Although some overlap exists between negative symptoms and social cognition in schizophrenia, according to participants at a National Institute of Mental Health (NIMH) conference where this issue was addressed [39] it is unwise to combine the constructs at this point in time. The consensus was that it is more informative to study negative symptoms and social cognition separately and to analyse relationships between them, until we know more about areas of convergence and divergence. Regarding positive symptoms [32], there has been some linkage between attributional style and paranoid delusions [40]. There is virtually no literature that has developed the relationship between disorganisation symptoms and social cognition.

Penn et al. [1] argued that multifactorial models of schizophrenia, including only non-social cognitive processes i.e. ‘neurocognition’, did not adequately explain the social functioning impairment in schizophrenia. Subsequently, social cognition was seen as a key domain for consideration during the first meeting of the NIMH-sponsored Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) initiative [27] and it was ultimately included as one of the seven domains represented in the MATRICS Consensus Cognitive Battery for clinical trials in schizophrenia [39].

Therefore, current accepted wisdom is that social cognitive deficit is a variety of neurocognitive compromise; it does not represent a symptom or group of symptoms by itself. The development of objective tests to quantify degrees of social cognitive impairment argues further for the validity of this construct.

Does Antipsychotic Treatment Improve Social Cognition?

Lewis and Garver [34] assessed facial affect recognition in 18 patients on haloperidol (5–20 mg/day) compared with 10 healthy controls in the course of their 2-week non-randomised clinical trial. An impairment in facial affect recognition was found in the schizophrenia group. This deficit was not related to psychopathology symptom scores.

Bellack et al. [41] assessed the effects of clozapine and risperidone on social skills at baseline, week 17, and week 29 in patients with schizophrenia using the Maryland Assessment of Social Competence. No significant medication effect on social competence was found despite clinical improvement on both medications. Similarly, Herbener et al. [31] described no beneficial effect of antipsychotics after 1 month (nine patients on risperidone) regarding facial affect recognition in 13 patients with first-episode psychosis.

It is worth noting that these studies were not adequately powered to draw definite conclusions. However, two influential randomised studies failed again to support the hypothesis that antipsychotics improve social cognition [29, 42]. Harvey et al. [29] found that patients with schizophrenia treated with either risperidone (n = 154; 2–8 mg/day) or quetiapine (n = 135; 200–800 mg/day) for an 8-week period of double-blind treatment did not improve their facial affect recognition, using the Penn Emotional Acuity Test. Similarly, Sergi et al. [42] found no evidence of treatment-related differences in social cognition in 73 patients with a diagnosis of schizophrenia in an 8-week double-blind study of risperidone, olanzapine, and haloperidol. Interestingly, when the potential influence of changes in neurocognition was statistically controlled for, there was no within-group change in social cognition. This suggests that social cognition and neurocognition are not the same thing, in other words, they vary independently of each other.

Alternatively, Gaebel and Wölwer [23] and Roberts et al. [38] demonstrated, respectively, significant improvement in facial affect recognition in patients on FGAs (haloperidol or perazine) and on the Social Cue Recognition Test in patients treated with olanzapine (n = 117) or quetiapine (n = 106) [Table 1]. Similarly Mizrahi et al. [21] studied 17 drug-free patients who then received antipsychotic treatment for 6 weeks: the effect on psychotic symptoms and ToM, using a hinting task, was measured every 2 weeks. The hinting task score was associated with negative and general symptom scores. Both the Positive and Negative Syndrome Scale (PANSS) positive scores and ToM improved after medication was started, particularly during the first 2 weeks of antipsychotic treatment.

Surprisingly, Machado de Sousa and Hallak [43] reported no differences in recognition accuracy or emotional intensity scores within the Facial Emotion Recognition Task between patients on clozapine compared with healthy controls. Since clozapine is a superior antipsychotic drug, this suggests that clozapine treatment may have corrected any deficit. The analysis of individual emotions, however, demonstrated a specific time-related deficit affecting the recognition of fear and disgust. Moreover, Harvey et al. [29] reported a similar apparent differential effect in patients treated for 8 weeks with quetiapine or risperidone: emotion perception remained unchanged, whereas social competence improved. This correlated with concurrent improvement in other aspects of neuropsychological performance, such as executive function and memory.

Is There Any Differential Effect on Social Cognition Between Antipsychotic Agents?

A number of influential studies have confirmed that SGAs outperformed FGAs in a range of clinical efficacy parameters, including the domain of social cognition [32, 36, 44–46]. It has been argued that SGAs’ strong affinity for 5-HT2 receptors [14] via the disinhibitory effect of serotonin antagonism on dopamine release in the prefrontal area may eventually improve emotion perception and social functioning [14]. However, both FGAs and SGAs also affect dopamine regulation in the mesocorticolimbic system, which suggests the potential for regulation of the amygdalae as an emotional manager [17].

Furthermore, clozapine and olanzapine increased dopamine outflow in the medial prefrontal cortex (mPFC), but not in the striatum or nucleus accumbens, whereas haloperidol had no effect in the mPFC but increased dopamine outflow in the striatum [47].

As a rule of thumb, frontal dopamine deficiency, perhaps as a response to striatal overactivity, has been considered germane to the induction of negative symptoms, associated with cognitive deficit and impaired social cognition. Therefore, trials have investigated the differential effects of FGAs and SGAs upon social cognition.

Kee et al. [32] evaluated the ability to recognise “emotional” faces in 20 treatment-resistant patients at baseline and after 8 weeks of treatment with risperidone or haloperidol, in a double-blind trial. The results of this study confirmed the positive influence of treatment with risperidone on the performance of facial affect tasks. Williams et al. [36] later reported similarly, that schizophrenic patients on haloperidol underperformed those on risperidone and healthy controls in recognising facial emotional expressions. Haloperidol-treated patients showed reduced fixation (attention) to salient features for neutral and happy expressions whereas risperidone-treated subjects and healthy controls achieved comparable results, displaying significantly better fixation to salient features for these expressions. This was followed by Littrell et al. [44] who in an open study found that 22 schizophrenia patients treated with olanzapine for 12 months performed better on a social perception measure, the Interpersonal Perception Task, than 30 patients on FGAs.

Fakra et al. [46] reported that 25 acute schizophrenia patients randomised to risperidone performed a facial affect discrimination task significantly better than those treated with haloperidol after 4 weeks. It was concluded that risperidone may specifically act on the processing of emotion-laden information: findings could not be explained on the grounds of facial recognition alone. However, Savina and Beninger [45] demonstrated that ToM performance in schizophrenia patients was related to maintenance, rather than acute treatment effects: they suggested that olanzapine and clozapine, but not risperidone or FGAs, may improve or protect ToM ability in this scenario.

It is worth mentioning that none of the studies above, apart from that of Harvey et al. [29], was both randomised and adequately powered. The lack of standardised social cognitive measures coupled with psychopathology rating scales detracts from their value. Common sense dictates that the active, distressing symptoms of acutely ill patients and the far from optimal state of arousal that these induce must seriously impair performance of any cognitive task that requires optimal attention and concentration. ‘Control’ tasks, to uphold the specificity of any improvement in social cognition, are conspicuous by their absence.

Consistent with the conclusion that these positive findings may be more apparent than real is a substantial body of literature reflecting far fewer differences between atypical and conventional antipsychotic drugs than initially suggested [48].

Of enormous influence is the CATIE trial (Clinical Antipsychotic Trials for Intervention Effectiveness trial), which failed to demonstrate differential antipsychotic effects on social cognition. To wit, Penn et al. [33] assessed emotion perception in 873 CATIE patients randomised to quetiapine, olanzapine, risperidone, ziprasidone (all SGAs), or perphenazine (FGA). Patients completed the Face Emotion Discrimination Task [49] immediately prior to randomisation and after 2 months of treatment. At baseline, 60 % of participants were on a SGA, 15 % on a FGA, and 25 % of subjects were antipsychotic free. Non-statistically significant improvement in emotion perception at 2 months was observed: the treatment groups did not differ from one another.

Finally, Kucharska-Pietura et al. [35] assessed deficits in social cognitive functioning in a naturalistic pragmatic sample of partially remitted stable schizophrenia inpatients, 28 being treated with a FGA (perphenazine or haloperidol), 56 being treated with a SGA (olanzapine or clozapine), and 50 healthy controls. In line with previous findings, there were no differences between the patient groups in emotional perception and ToM/empathy. This is particularly striking given the supposedly superior effects of clozapine previously reported. There were small but significant advantages for SGAs in non-social low-level visual processing: this was thought to result from SGAs’ weaker antagonism of dopamine receptors in the retina [35].

Conclusions

First, overall, antipsychotic drugs of either class demonstrate little reliable effect upon social cognition [38, 50]. There is a modicum of support for the use of oxytocin as an adjunct to antipsychotic drugs [11] but whether this latest finding is a valid effect remains a matter of conjecture. By contrast, recent randomized intervention studies of specialised psychosocial treatment programmes for social cognition report very promising results in the improvement of emotional perception and social skills in schizophrenia [50-52].

Secondly, the literature suffers from inconsistencies in study design, particularly a prevalence of non-randomised approaches based upon cross-sectional assessments, which do not reflect the later NIMH recommendations. Nor are medication doses standardised. Most sample sizes are quite small, and there is inadequate control of pertinent clinical variables. This overlaps with three obstacles to research progress identified by the NIMH group: (1) psychometrics and measurement, (2) maturity of the field, and (3) a lack of interdisciplinary bridges between clinical and basic researchers [39].

Finally, large-scale longitudinal investigations are needed to explore the unclear relationships between social cognition, symptoms, and functional outcome. If social cognition proves to represent a neurocognitive construct, we suspect related to premorbid personality, then it is not logical to expect current antipsychotic treatments designed to attenuate active symptoms to have any significant effect other than through symptom control, thus abolishing the ‘noise’ of symptoms in the patient’s attempts at social cognitive function. Other treatments, quite possibly training patients in the areas in which they are impaired, may hold more promise.