Schizophrenia is an important psychiatric disorder characterised by different psychopathological dimensions such as positive, negative or cognitive symptoms. A key component of cognitive functions is cognitive control, which has proven to be essential for the proper execution of goal maintenance, set-shifting and inhibition.Reference Botvinick, Carter, Braver, Barch and Cohen1 Cognitive control is necessary when we block an habitual or automatic behaviour, and instead execute a less familiar/alternative behaviour.Reference Matsumoto and Tanaka2 It is essential for daily life and its deficit has been associated with social and behavioural disturbances.Reference Larson, South, Clayson and Clawson3 Apart from deficits in neurocognition, schizophrenia usually involves impairments in emotional processing, a domain of social cognition abilities.Reference Green, Waldron and Coltheart4–Reference Patrick, Rastogi and Christensen6 Particularly, the ability to understand emotional cues and integrating emotion perception with context is severely disturbed.Reference Kring and Elis7,Reference Romero-Ferreiro, Aguado, Torío, Sánchez-Morla, Caballero-González and Rodriguez-Jimenez8 Facing conflicting cues is relatively common in social interactions, and can happen when, for example, someone masks their true feelings or there is sarcastic irony. These conflicting cues present an even greater challenge for patients with schizophrenia.Reference Mitchell and Rossell9 Consequences of cognitive and emotional processing deficits have been studied separately, and the possible common links behind those deficits have not been extensively described.Reference Anticevic and Corlett10,Reference Ferrer-Quintero, Fernández, López-Carrilero, Birulés, Barajas and Lorente-Rovira11 The search for a greater understanding of the association between executive function and emotion has motivated the use of experimental paradigms such as the emotional version of the Stroop task.Reference Stroop12,Reference Williams, Mathews and MacLeod13 In this version, the emotional valence of the word usually slows down the identification of the ink colour, suggesting that the affective content of stimuli affects the processing of perceptual aspects of it.Reference Burt14 The emotional Stroop task has been proposed as a useful task to link several abnormalities in the cognitive functioning of schizophrenia as a neuropsychological endophenotype.Reference Aleksandrowicz, Hagenmuller, Haker, Heekeren, Theodoridou and Walitza15,Reference Feroz, Leicht, Rauh and Mulert16 It combines emotional (as distractor) and non-emotional (as required response) information (e.g. naming the ink colour of an emotional word). Only a few attempts of comparing patients with schizophrenia in a cognitive versus emotional Stroop task involving faces have been made. They have found slower response times in incongruent compared with congruent trials in both versions of the task.Reference Egner, Etkin, Gale and Hirsch17,Reference Hurtado, Triviño, Panadero, Arnedo and Tudela18 However, in these studies, the cognitive task was to indicate the gender of a face with an incongruent word (the contrary gender) superimposed.
The main aim of this research is to study cognitive control understood as the capacity of patients to suppress irrelevant information in cognitive-emotional versus purely emotional versions of the emotional Stroop task. Although both versions included emotional words as distractors, the attribute required from the participant could be either emotional (i.e. the emotional valence of a face) or perceptual (i.e. the ink colour of the word). This procedure allows to study to what extent cognitive and emotional control are affected in patients with schizophrenia. As a secondary objective, the relationship between performance on both types of Stroop tasks and some clinical variables (symptoms, medication dosage and age at onset of the disorder) was examined.
Method
Participants
Twenty-nine in-patients with schizophrenia were included in this study. Patients were recruited at the acute psychiatric unit of the Hospital Universitario 12 de Octubre (Madrid, Spain). They took part in this study at the time of hospital discharge, when their referring psychiatrist felt that they had a certain degree of psychopathological stability. All patients were diagnosed with schizophrenia according to DSM-5 criteria,19 using the Structured Clinical Interview for DSM-5, Research Version.Reference First, Williams, Karg and Spitzer20 All participants were on second-generation antipsychotics, except for four participants who were receiving a combination of first- and second-generation antipsychotics, and two participants who were on first-generation antipsychotics only. Additionally, 19 participants were receiving another type of psychotropic drug treatment, 16 of which were benzodiazepines. Antipsychotic treatment was converted to chlorpromazine equivalents.Reference Leucht, Samara, Heres and Davis21 Clinical status was evaluated with the Spanish version of the Positive and Negative Syndrome Scale (PANSS).Reference Kay, Fiszbein and Opfer22,Reference Peralta and Cuesta23 On that basis, the Wallwork five-factor model for the PANSS (positive, negative, cognitive/disorganised, excited and depressed), which has been validated in the Spanish population, was also calculated.Reference Wallwork, Fortgang, Hashimoto, Weinberger and Dickinson24,Reference Rodriguez-Jimenez, Bagney, Mezquita, Martinez-Gras, Sanchez-Morla and Mesa25 See Table 1 for sociodemographic variables and clinical status.
Table 1 Sample demographics and clinical status of patients
The consensus five-factor Wallwork model (for the PANSS) scores were calculated. PANSS, Positive and Negative Syndrome Scale; CPZeq, chlorpromazine equivalent dose (mg/day).
The inclusion criteria were as follows: diagnosis of schizophrenia disorder according to DSM-5 criteria, clinical stabilisation before discharge from the hospital unit, aged 18–55 years and sufficient fluency in Spanish to allow them to complete the protocol. Exclusion criteria for patients included the presence of electroconvulsive therapy in the previous year, neurological disorders or somatic diseases that could interfere with the performance of the tasks, any active major substance misuse/dependence (excluding caffeine and nicotine), intellectual disability (IQ<70) evaluated with the Spanish version of the Word Accentuation Test,Reference Gomar, Ortiz-Gil, McKenna, Salvador, Sans-Sansa and Sarró26 autism spectrum disorder, inability or unwillingness to participate and other major psychiatric comorbidities such as affective disorders. The comparison group was composed of 29 matched (in terms of age and gender) controls with no personal or first-degree family history of psychotic disorders.
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. All procedures were approved by the Ethics Committee of the Hospital Universitario 12 de Octubre (approval number 17/308). Written informed consent was obtained from all participants before their inclusion in the study.
Materials
The present study comprised two tasks developed within a Stroop paradigm framework. The experimental design was specifically developed for this purpose. Although both versions of the task included emotional stimuli, one of them required a response about the emotional valence of the face (which is why it is called the emotional-emotional Stroop task (EESt)), whereas the other task required perceptual aspects of the stimuli, i.e. the ink colour of an emotional word (cognitive-emotional Stroop task (CESt)).
The target facial stimuli consisted of 36 pictures taken from the NimStim stimulus set.Reference Tottenham, Tanaka, Leon, Mccarry, Nurse and Hare27 Three pictures from each model, showing happy, angry and neutral expressions, were selected. Additionally, 72 emotional words (24 positive, 24 negative and 24 neutral) were selected from two different databases.Reference Hinojosa, Martínez-García, Villalba-García, Fernández-Folgueiras, Sánchez-Carmona and Pozo28,Reference Redondo, Fraga, Padrón and Comesaña29 The length of the words and frequency of use were controlled to remain constant along the three emotional categories. All words had between three and four syllables and ≥30 points in relative frequency of use per million. Stimuli were selected based on their valence and arousal values. Words with extreme valence scores (either positive or negative) were selected, ensuring that they had similar arousal values and the smallest possible s.d. The mean valence value of the final set was 7.54 (s.d. = 0.46) for positive words and 2.32 (s.d. = 0.73) for negative words (t(46) = 41.98, P < 0.001), on a scale with anchor points from 1 (very negative) to 9 (very positive). Positive and negative words did not differ in terms of arousal (t(46) = 1.29, P = 0.203; positive pictures: mean 6.06, s.d. = 0.36; negative pictures: mean 6.08, s.d. = 0.99). The arousal scale ranged from 1 (very calming) to 9 (very arousing). The neutral words had a mean valence of 4.97 (s.d. = 0.27) and mean arousal of 4.03 (s.d. = 0.39).
Procedure
The procedure was developed in E-Prime version 2.0 for Windows (PsychologySoftware Tools, Pittsburgh, USA; https://pstnet.com/products/e-prime/). For the EESt, a target face appeared with an emotional word superimposed. Participants were asked to categorise the emotional valence of the face while ignoring the word. Keyboard responses were given by tapping buttons 1, 2 or 3, corresponding to positive, neutral or negative valence. Participants were free to use their preferred hand. Figure 1 shows an example of the screen response. In this task, words were randomly divided into three groups, each one with the same number of positive, negative and neutral words. Each group was randomly paired with positive, negative and neutral faces in such a way that all of the experimental conditions had the same number of stimuli.
Fig. 1 (a) Example of a trial of the EESt. Participants had to respond to the emotional valence of the face, ignoring the word. (b) Example of a trial of the CESt. Participants had to respond to the ink colour of the word, ignoring its meaning. CESt, cognitive-emotional Stroop task; EESt, emotional-emotional Stroop task.
Then, the second block containing the cognitive version of the task (CESt) was administered. Participants were presented with a target word (the same set presented in the EESt) printed in either red, blue or green. They were asked to select the colour of the word while ignoring its meaning. As in the previous task, each category of words (positive, negative and neutral) was divided into three subsets, with an equal number of stimuli. Figure 1(b) shows a screenshot of this task. Accuracy data as well as response times were recorded for both tasks. The final number of trials for both tasks was 72. The order of blocks was counterbalanced across participants.
Data analysis
First, a priori sample size was calculated with G*Power for Windows (Heinrich Heine University, Düsseldorf, Germany; https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-und-arbeitspsychologie/gpower) version 3.1. We wanted to be able to detect a relatively small effect size of η 2p = 0.4, using an alpha level of 0.05. The total sample size resulted in 54 participants, which would give a statistical power of 80%. Altogether, we included 58 participants.
Data were managed and analysed with IBM SPSS for Windows (IBM Corporation, Chicago, USA; https://www.ibm.com/spss) version 24. Simple comparisons (χ²-test for gender distribution and t-tests for age and years of education) between patients and controls were used for sociodemographic variables.
Regarding analysis of response times, we first removed the error trials before performing the computing analysis. Then, positive and negative interference scores were calculated. Interference is, by definition, a difference score between a baseline or non-conflicting measure (in this case, those trials in which the word was neutral) and a conflict condition (e.g. a trial in which the face was happy and the word was negative). Consequently, the negative interference score was calculated by subtracting the average response time of trials where the distractor was a neutral word, from negative incongruent trials. The same procedure was followed for calculating the positive interference score. Once interference effect was calculated for each condition, a 2 × 2 × 2 repeated-measures analysis of variance was conducted. The task (EESt, CESt) and the interference (positive, negative) were the within-participant factors, whereas the group (patients, controls) was the between-participants factor. Greenhouse–Geisser and Bonferroni corrections were applied when appropriate. Finally, Pearson's correlations were conducted to explore the possible association of performance in both tasks with clinical symptoms (PANSS factors according to the Wallwork model) and antipsychotic medication dosage (chlorpromazine equivalents). Results were considered significant when P < 0.05.
Results
As Table 1 shows, patients did not differ from controls in terms of age (t(56) = 0.369, P = 0.71) or gender distribution (χ2 = 1.73, P = 0.18). Patients had significantly fewer years of education compared with healthy controls (t(51) = 7.61, P < 0.001). Mean response times and error rates for each of the conditions in the task are presented in Table 2.
Table 2 Mean response times (milliseconds) and error rates for each group and experimental condition of each task
Interference effect (response times)
Here, both tasks were submitted to the same analysis. First, a main effect of group was found (F(1,40) = 9.44, P = 0.004, η 2p = 0.19). Patients showed higher interference than controls (patients: mean 393.33, s.d. = 425.37; controls: mean 90.72, s.d. = 317.08; 95% CI 103.51–501.73). Second, a task×group interaction was found (F(1,40) = 12.18, P = 0.001, η 2p = 0.23). Although there were no significant differences between patients and controls in the CESt (P = 0.831), the interference in the EESt was higher for patients than for controls (patients: mean 744.19, s.d. = 738.36; controls: mean 152.39, s.d. = 550.31; P = 0.001, 95% CI 246.20–937.42). Additionally, an interference×group interaction was also present (F(1,40) = 7.705, P = 0.008, η 2p = 0.16). Regarding negative interference, there were no differences between patients and controls (P = 0.418). However, patients with schizophrenia showed higher interference effects than controls in positive incongruent trials (patients: mean 649.63, s.d. = 563.07; controls: mean 139.77, s.d. = 419.67; P < 0.001, 95% CI 246.30–773.43). Finally, the triple interaction task×interference×group was marginally significant (F(1,40) = 2.91, P = 0.096, η 2p = 0.07) (see Fig. 2). Similarly, post hoc comparisons revealed that in the EESt, patients showed greater positive interference than controls (patients: mean 1189.92, s.d. = 1016.4; controls: mean 269.11, s.d. = 757.6; P < 0.001, 95% CI 445.04–1396.58).
Fig. 2 Interference effect separated by affective valence of distractor words and task. Error bars denote s.d. CESt, cognitive-emotional Stroop task; EESt, emotional-emotional Stroop task. **P < 0.001.
Cross-task and clinical variables correlations
The correlation between clinical variables and positive and negative interference in both the EESt and CESt are shown in Table 3. As can be seen, no significant correlations were found between tasks and symptoms, medication dose or age at onset of the disorder, except for the EESt negative interference condition and the cognitive/disorganised factor of the Wallwork model (r = 0.46, P = 0.04).
Table 3 Pearson's correlations [95% confidence intervals] between interference in the cognitive-emotional Stroop task and emotional-emotional Stroop task, age at onset of disorder, chlorpromazine equivalent dose and the consensus five-factor Wallwork model (for the PANSS) scores, for patients diagnosed with schizophrenia
CPZeq, chlorpromazine equivalent dose (mg/day); PANSS, Positive and Negative Syndrome Scale.
*P < 0.05; significant correlations are highlighted in bold.
Discussion
The main objective of this study was to compare a group of patients with schizophrenia with a healthy control group in a cognitive versus emotional conflict modulation task. For this purpose, two emotional versions based on the Stroop paradigm were designed. In general terms, results reported in this study support the idea that patients with schizophrenia show a more pronounced deficit in emotional as opposed to cognitive control, compared with a group of healthy controls. Specifically, that deficit appears to be slanted toward positive emotional stimuli.
Interference measures based on response times suggested a poorer capacity of patients to inhibit irrelevant emotional information in the emotional conflict task compared with that of healthy controls. This outcome supports the notion that emotional information selectively interferes with the processing of emotional aspects of the target stimuli to a high degree in patients with schizophrenia. Analogous to the current cognitive task, previous literature found overall slower response times for patients with schizophrenia compared with controls.Reference Demily, Attala, Fouldrin, Czernecki, Ménard and Lamy30,Reference Sollier-Guillery, Fortier, Dondaine, Batail, Robert and Drapier31 Nevertheless, our results give additional evidence in the context of a within-task comparison, in which differences between patients and controls in the cognitive task appear to be smaller than those found in the emotional task. One possible explanation for the longer response times in patients with schizophrenia might be related to proactive versus reactive modes of cognitive control. Proactive control relies on the anticipation and prevention of interference before it occurs, whereas reactive control is considered a ‘late correction’ mechanism that is mobilised when a high-interference event is detected.Reference Barch and Ceaser32 Thus, the reactive form of control is less effective. Some authors have suggested that proactive control may be particularly impaired schizophrenia.Reference Edwards, Barch and Braver33 This would lead individuals with schizophrenia to rely more heavily on reactive control, especially when emotion-related conflicting information is handled.
Of special interest is the greater interference effect in the emotional task when the emotional valence of the word was positive found in the group of patients with schizophrenia. One possible explanation is that identifying the target emotion of the face itself was more difficult for patients in this experimental condition compared with identifying the expression of a neutral or a happy face in the negative interference condition. This could be a plausible explanation considering that the most frequently found pattern in literature is a preserved recognition of positive expressions and an impaired recognition of negative emotions.Reference Kohler, Turner, Bilker, Brensinger, Siegel and Kanes34,Reference Romero-Ferreiro, Aguado, Rodriguez-Torresano, Palomo and Rodriguez-Jimenez35 Nevertheless, this effect should be treated with caution as the triple interaction was only marginally significant. Further studies are necessary to confirm this preliminary outcome.
As a secondary objective, the relationship between performance of patients in both tasks and some clinical variables was studied. In this regard, no significant correlations with positive or negative PANSS factors, antipsychotic dosage or age at onset of disorder were found. Such results can be interpreted as a deficit, which is present in patients with schizophrenia regardless of the classic symptom course. The absence of a correlation with negative symptoms could be striking, given the frequent overlapping between negative and cognitive symptoms and the link with neurobiological impairments such as the limbic system,Reference Fernandez-Egea, Parellada, Sugranyes, Horga, Lomeña and Falcon36 or even inflammatory conditions.Reference Goldsmith and Rapaport37 Nevertheless, other authors have also found cognitive deficits to be independent of positive and negative symptoms in schizophrenia.Reference Heaton, Gladsjo, Palmer, Kuck, Marcotte and Jeste38 Despite this, a positive correlation between the negative interference in the emotional version of the Stroop task and the cognitive/disorganised PANSS factor was found. This finding makes sense, considering that this factor of the PANSS evaluates clinically cognitive impairments in a broader sense.Reference Rodriguez-Jimenez, Bagney, Mezquita, Martinez-Gras, Sanchez-Morla and Mesa25 Some previous studies have found a relationship between cognitive/disorganised symptoms and a variety of tasks evaluating inhibitory control in schizophrenia.Reference Ngan and Liddle39,Reference Moritz, Andresen, Jacobsen, Mersmann, Wilke and Lambert40 It is possible that because threatening information has the greatest capacity to capture attention,Reference Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg and Van Ijzendoorn41 this could explain the correlation between the cognitive/disorganised PANSS factor and the specific condition of negative interference. In this sense, those patients with symptoms of disorganisation would find it harder to redirect attention to the target compared with patients with other clinical profiles or control participants. From the clinical point of view, this result highlights the importance of cognitive control as a useful target for intervention regardless of the predominant symptoms in schizophrenia. In this regard, some results indicate that cognitive remediations aimed to increase response inhibition or executive function also had an impact on functional ability.Reference Tan, Zhu, Fan, Tan, Yang and Wang42 It remains to be seen whether specific training of emotion-related conflict has an impact on ability to understand emotional cues in schizophrenia.
The main strength of this study is the between-group experimental design, in which patients with schizophrenia and healthy controls were compared across two versions of the emotional Stroop task. The experimental design is rigorous, considering that the stimuli was selected from normative studies and matched in terms of arousal. Nevertheless, our study has some limitations. First, the sample included in-patients at discharge, and results may not be entirely generalisable to out-patients with schizophrenia. Also, generalisation of our results should be taken with caution, as patients with schizophrenia often have other major psychiatric comorbidities, but it was considered an exclusion criterion of our sample. Second, this paradigm creates a highly artificial situation not easily comparable with real-life situations. Finally, a more detailed clinical, as well as cognitive, characterisation of the patient sample would have been desirable.
In conclusion, patients with schizophrenia showed a decreased ability to inhibit the emotional valence of words, especially when competing information shared emotional content. These results help to provide a better understanding of the altered cognitive and emotional mechanisms in schizophrenia, and will be of value in the development of future cognitive rehabilitation programmes.
Data availability
The data that support the findings of this study are available from the corresponding author, V.R.-F., upon reasonable request.
Author contributions
V.R.-F., A.G.-G. and R.R.-J. designed the study. A.G.-G., P.M.-B. and J.A.P. outlined the methodology. V.R.-F., P.R.-G. and C.R. undertook the statistical analysis. V.R.-F., A.G.-G., I.T., M.A.A.-M. collected data. V.R.-F., A.G.-G., I.T. and R.R.-J. wrote the first draft of the manuscript. P.M.-B., E.M.M. and R.R.-J. substantially reviewed the manuscript. All authors have read and approved the submitted version, and agree to account for their own contributions and the accuracy or integrity of any part of the work.
Funding
This work was supported in part by the Spanish Ministerio de Ciencia, Innovación y Universidades (grant number PSI2018-098876-B-100) to E.M.M., and in part by the Instituto de Salud Carlos III (grant numbers PI16/00359 and PI19/00766; Fondo de Investigaciones Sanitarias/FEDER), the European Development Regional Fund ‘A way to achieve Europe’ (ERDF), the Madrid Regional Government (R&D Activities in Biomedicine S2017/BMD-3740 (AGES-CM 2-CM)) and Structural Funds of the European Union.
Declaration of interest
R.R.-J. has been a consultant for, spoken in activities of, or received grants from the Instituto de Salud Carlos III, Fondo de Investigación Sanitaria (FIS), Centro de Investigación Biomédica en Red de Salud Mental (CIBERSAM), Madrid Regional Government (S2010/BMD-2422 AGES; S2017/BMD-3740), Janssen-Cilag, Lundbeck, Otsuka, Pfizer, Ferrer, Juste, Takeda, Exeltis, Angelini, Casen-Recordati and Rovi. The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Open access
eLetters
No eLetters have been published for this article.