Social Perception in Mesial Temporal Lobe Epilepsy: Interpreting Social Information From Moving Shapes and Biological Motion

Epilepsy is a CNS disorder affecting about 70 million people worldwide.¹ The most common type is mesial temporal lobe epilepsy (MTLE), characterized by focal seizures usually associated with lesions (hippocampal sclerosis or focal cortical dysplasia) in the mesial temporal zone. As MTLE is often drug resistant, the treatment of choice is surgery of the epileptogenic focus.² One of the main neurosurgical approaches with regard to seizure outcome is the anterior temporal lobectomy (ATL). ATL comprises the resection of anterior parts of the temporal lobe, including the amygdala, hippocampus, anterior part of the fusiform gyrus and adjacent neocortical temporal tissue.³ Both the MTLE and its surgical treatment may affect the functioning of patients by changing the structure and connectivity of the temporo-limbic and frontal areas.⁴

Recent research has suggested that impairments in the ability to correctly interpret and act upon the behavior and mental states of other people may be at the root of social disturbances observed in MTLE.⁵ Some may argue that these deficits in social cognition can also be a result of psychosocial aspects of epilepsy like prolonged stress, alienation, stigma, discrimination, or antiepileptic drugs. Nevertheless, it must be noted that those difficulties occur more frequently in patients with MTLE compared with those with other types of epilepsy. This suggests that neuronal abnormalities in the mesial temporal area may lead to deficits in social cognition in this clinical group.^4,6

Social cognition is an umbrella term for a wide range of processes associated with processing of social information, from detection and identification of the social agents (social perception) through the processing of their affective states (emotion processing) to the complex inference of mental states, goals, or beliefs (theory of mind).⁷ Each of these abilities has been associated with the activity of widespread brain networks: for example, fusiform areas, superior temporal sulcus, and frontal mirror network system can be linked to the person’s perception, while temporo-parietal junction and medial prefrontal cortex activity is usually found during activities engaging theory of mind processes.⁸ Furthermore, both neuroimaging⁹ and lesion¹⁰ evidence strongly points to the crucial role of medial temporal system (particularly amygdalar) activity for emotion processing.

In line with these observations, researchers¹¹ observed that theory of mind deficits can occur in patients with focal seizures emanating from the temporal and frontal lobes, but not in individuals with focal seizures originating from other structures. While emotion processing^12,13 and theory of mind^14,15 skills have previously been examined in patients with epilepsy, assessment of other domains of social cognition as identified by the National Institute of Mental Health Consensus Statement⁷(social perception and knowledge, attributional style) are scarce. Some domains worth mentioning are processing of nonverbal cues for the understanding of individuals’ social roles (social perception) or awareness of rules that guide human interactions (social knowledge). As noted by several researchers,⁵ the comprehensive evaluation of social cognitive skills in laboratory conditions may be a challenging task. However, it is necessary, given the detrimental effect of these impairments on patients’ lives.

The main aim of our research is a multilevel evaluation of social cognition skills in participants with MTLE. Humans attribute social behavior to motion observed at a very low level of complexity (e.g., simple geometric forms moving in relation to one another), as well as at a higher level (e.g., humans performing tasks or communicating with each other).¹⁶ We decided to rely on tasks measuring social inference from both kinds of motion: biological motion displays of human communicative interactions (Communicative Interactions Database-5 Alternative Forced Choice format [CID–5])¹⁷ and motion performed by geometric shapes (FHA).

According to previous studies, perception of body movement significantly complements information originating from facial expression.^18,19 Correctly interpreting other individuals’ bodily behaviors (posture, dynamics of movement, meaningful coordination of actions) is an important skill and has been shown to lead to the experience of more meaningful relationships, as well as social approval.^19,20 At the same time, previous studies on social cognition in MTLE^5,21 were based on highly complex vignettes presenting social situations with real-life actors (e.g., The Awareness of Social Inference Test [TASIT]²²). This approach, while presenting high ecological validity, may be limited by the fact that correct interpretation of the vignettes is based on a number of factors that must be taken into account while processing actions of the agents (e.g., facial expressions, gaze direction, verbal prosody, body movements, etc.). The use of the point-light stimuli and/or geometric shapes allows one to study processing of communicative interaction from movement of the agents, while limiting the amount of visual input and minimizing the impact of the other factors (e.g., deficient face or prosody processing) on the patients’ performance. Our particular interest is to determine whether point-light stimuli reflecting human biological movement can be used to further assess social cognition in MTLE patients and their ability to extract social information from their environment. To the best of our knowledge, it is the first study to use body motion assessment to investigate social cognitive processes in patients with epilepsy with a paradigm that previously has been effectively used to investigate social cognitive processes in patients with schizophrenia²³ and adult patients with autism spectrum disorders.²⁴

Here, we compare three groups of participants: patients with intractable MTLE with ATL (ATL+), patients with intractable MTLE without ATL (ATL–), and healthy control subjects. Additionally, we investigate whether clinical or demographic variables are related to any aspect of social cognition.

Methods

Participants

For this case-control study, we recruited 21 patients with drug-resistant MTLE without lobectomy and 19 demographically matched patients following lobectomy during their diagnostic hospitalization in the neurosurgery ward.

Neurological diagnostics determining the localization of the epileptic focus included video-EEG, MRI, and positron emission tomography. Moreover, all candidates for surgery underwent the Wada Test as the last stage of the medical evaluation. During the surgery, electrocorticography monitoring was used to determine the exact area of the epileptogenic zone to be removed. All patients underwent a full neuropsychological examination.

Exclusion criteria were history of any other neurological or psychiatric treatment, drug or alcohol addiction, intellectual impairment, verbal dysfunction, and physical disabilities that would impair patients’ ability to complete the task properly (e.g., major visual deficit, paresis of the dominant hand). Demographically matched (in pairs: case-control) healthy participants were recruited from volunteers without any declared history of psychiatric or neurological disorders, originating from the same community sample that responded to online advertisements. Demographic and clinical characteristics of the study participants are presented in Table 1, and details pertaining to histopathology and pharmacotherapy of the study sample are summarized in Table 2. Participants received no financial compensation for participation in the study.

Procedure

Each participant was examined individually during a single session. The duration of sessions varied between participants, depending on their neurological and physical condition, and lasted between 45 and 60 minutes. The whole procedure took place in a quiet room in the hospital during the morning hours.

At the beginning of a session, participants were asked to provide demographical information about themselves. Clinical data including details about the type, course, and treatment of epilepsy were obtained from patients’ medical charts and are presented in Table 1.

To examine whether social cognitive problems in our MTLE groups might be attributed to the general psychomotor slowing or executive deficits, we used well-established neuropsychological measures to assess basic psychomotor speed (Digit Symbol from WAIS-R^25,26) and executive functions/attentional control (Trail Making Test, Part B²⁷) for all participants.

The social cognition testing was completed by each participant in the same order. It comprised the FHA and the Communicative Interactions Database-5 Alternative Forced Choice format (CID-5) task. Procedure details for every task are described below.

All procedures were carried out in accordance with the Declaration of Helsinki and were approved by the ethics committee of the Faculty of Psychology at the University of Warsaw. All participants gave written informed consent to participate in the study.

Assessment

FHA.

The FHA²⁸ were used to evaluate participants’ ability to accurately attribute mental states and emotions to dynamic visual stimuli. Stimuli consisted of 12 animations (with three additional training animations at the beginning) featuring two moving triangles (a big red triangle and a small blue triangle) on a white background. Each animation lasted from 34 to 45 seconds and was shown on a 17-inch computer screen. The participants’ task was to describe the movement of the triangle while watching an animation, then to categorize it as either random motion (no evidence for any level of intentionality of movement of triangles), goal directed (when the triangle movement appeared goal directed and interdependent; e.g., one triangle is following another), or mentalizing (triangles engaging in mutual activity associated with manipulating each other’s thoughts, acts, and emotions; e.g., that one triangle is tricking another into leaving the room it is in by knocking at the door). Additionally, after animations depicting mentalizing, participants were asked to select the one out of five alternatives that best describes each triangle’s emotion (e.g., if the movement of a triangle resembles bouncing up and down or a dance, one could call it “cheerful”). Each set of alternatives included a “no feelings” answer as well as labels of four emotions (both negative and positive), which varied between the trials. A full list of the alternatives is provided elsewhere.²⁸ (Exemplary stimuli from each condition can be found online at https://sites.google.com/site/utafrith/research.) Participants could score between 0 and 12 points in interaction-type classification and between 0 and 8 points in emotion recognition. No time limit was given. Previous studies have documented that HCs are able to easily categorize FHA (scores ranging from 75+/−15% for goal directed to 98+/−7% for theory of mind items).²⁹ Furthermore, it has been documented that both the FHA categorization score and the FHA emotion recognition score effectively discriminate healthy adults from adults with autism spectrum disorders. Weak correlations were also found in adults with autism spectrum disorders between FHA categorization and a well-established verbal theory of mind task (false-belief task).²⁸

CID-5.

The CID-5 task aimed to provide information about the participants’ ability to interpret biological motion in terms of communicative intention or lack thereof.¹⁷ Stimuli consisted of 21 animations from the Communicative Interactions Database,³⁰ featuring two point-light agents—white on a black background—shown on a computer screen. In 14 animations, the agents were interacting (e.g., agent A squats down and asks agent B to imitate him; agent B squats down), while in seven animations they were acting independently of each other (e.g., agent A looks under his foot, agent B moves an object away). The order of animations was randomized. Each animation was played twice in a row, presenting the same stimuli from different angles (respectively 90 and 125 degrees). The participants’ task was to classify agents’ moves as communicative (when the action of one agent influenced the other agent) or individual (when the behavior of one agent did not impact the behavior of the other agent) and then choose the most appropriate description of agents’ actions from five possible options, which were different for each item. For example, one set of descriptions was: 1) agent A squats down, and asks agent B to imitate him; agent B squats down; 2) agent A squats down, and agent B takes something from the floor; and 3) agent A sits on the floor, and agent B bends down. (Stimuli used in the study can be found online at https://link.springer.com/article/10.3758%2FBRM.42.1.168.) No time limit was given.

The Polish version of the CID-5 task has been recently validated among six other language versions of the Multilingual CID-5 task.¹⁷ Similar to other versions of the CID-5 task, high accuracy for both interaction classification (89+/−9%) and recognition of specific actions (74+/−12%) was found in HCs. Furthermore, CID-5 scores have been shown to discriminate HCs from patients with schizophrenia²³ but not from high-functioning individuals with autism.²⁴

Statistical Analyses

Statistical analyses were performed using SPSS Statistics v.22 for Windows (IBM Corp., Armonk, N.Y.). One-way analysis of variance was used to investigate differences between groups in test scores. Dunnett’s test was used for post hoc comparisons. Spearman’s rho was used for the correlational analyses between tests scores and cognitive and clinical variables, due to the nonnormality of distribution of some variables (Kolmogorov-Smirnov, p<0.05).

The threshold for significance for correlational analyses was adjusted according to the number of correlations calculated for each type of variable. For demographic variables, five outcomes were correlated with two demographic scores (age, years of education), so the threshold was divided by 10, thus producing a new p=0.005. Similar treatment was used for the threshold for correlation between clinical (years of epilepsy, age of onset, number of seizures per month) and cognitive (Digit Symbol and Trail Making Test, Part B) scores.

Results

Participants

Groups of participants were matched in terms of age and gender distribution. HCs had more years of education than clinical groups. Clinical groups did not differ with respect to age of epilepsy onset or duration of illness. Patients who underwent surgery had fewer seizures than the group of patients without ATL. Detailed information on demographic and clinical variables is presented in Table 1.

Social Perception Tasks

Basic descriptive statistics for the results of FHA and CID-5 are presented in Table 3. No significant differences were found between groups with and without lobectomy for any of the social perception related variables.

FHA.

The main effect of the group was observed both for classification of the animation type and for emotion recognition. Further investigation revealed differences for goal directed and mentalizing but not random motion animations. Post hoc analysis revealed significantly worse overall FHA categorization in participants with (p=0.000) and without (p=0.000) lobectomy compared with HCs. For FHA emotion recognition, performance was worse for participants after lobectomy (p=0.04) compared with the healthy control group. Both without-lobectomy (p=0.003) and with-lobectomy (p=0.012) groups were less efficient in the recognition of goal-directed animations compared with HCs. Similarly, both without-lobectomy (p=0.01) and with-lobectomy (p=0.001) groups were less efficient in recognizing mentalizing animations than HCs.

All of these effects remained significant after inclusion of years of education as a covariate in the analysis (FHA categorization: F=5.7 df=2, 62, p=0.005; FHA goal-directed: F=3.3, df=2, 62, p=0.043; FHA mentalizing: F=4.6, df=2, 62, p=0.014; FHA emotion recognition: F=4.0, df=2, 62, p=0.023).

CID-5 task.

A main effect of the group was observed for the recognition of interactions between point-light agents. Without-lobectomy patients performed significantly worse than controls in interaction recognition (Dunnet’s p=0.011).

No difference between groups was found for communication condition recognition. However, significant differences were found for CID-5 individual condition scores. Both without-lobectomy (p=0.002) and with-lobectomy (p=0.001) groups classified individual items as communicative more often than HCs. Both ATL– (p=0.047) and ATL+ (p=0.003) groups were less efficient at naming the correct individual actions in the forced-choice task than HCs.

Side of Epileptic Focus

In the ATL– group, no significant differences were found between patients with left- or right-sided epileptic focus. However, we found one such difference in ATL+ patients. In the CID-5 test, patients with right-sided epilepsy made more mistakes in choosing the proper description of communicating agents’ actions (p=0.015).

Demographic Correlations

Subjects who had more years of education received higher scores in both FHA categorization (rho=0.46, p=0.002) and CID-5 forced-choice recognition (rho=0.49, p=0.001).

Clinical.

Longer time of epilepsy duration (in years) was related to lower FHA categorization scores (rho=−0.47, p=0.002). Patients with a higher mean number of seizures had lower scores on the CID-5 agent interaction identification (rho=−0.47, p=0.003).

Cognitive.

Longer time of performance on the Trail Making Test, Part B was associated with lower CID-5 forced-choice recognition (among five responses; rho=−0.53, p=0.001). Higher Digit Symbol score was associated with better CID-5 classification (rho=0.45, p=0.005) and forced-choice recognition (rho=0.48, p=0.002).

Discussion

The main aim of this study was to examine social perception processes in patients with mesial temporal lobe epilepsy with or without surgical treatment. We extended the previous findings of deficient social inference abilities in epilepsy^5,6,13,15,21 by showing that irrespective of being subjected to lobectomy, patients with MTLE display problems with the interpretation of social actions presented with both biological (CID-5) and animate (FHA) motion. Both with- and without-lobectomy patients displayed deficient recognition of individual actions presented with point-light action in CID-5 and decreased ability to recognize goal-directed and mentalizing animations during FHA paradigm. Interestingly, without-lobectomy patients did not differ from controls in any forced-choice tasks (CID-5 forced-choice, FHA emotion recognition) but were outperformed by controls in CID-5 interaction recognition. Conversely, patients with ATL were less able than controls to choose the correct description of individual items in CID-5 and to correctly recognize specific emotion of the agents in the FHA.

To the best of our knowledge, this is the first study to use point-light motion displays to investigate social cognitive processes in patients with epilepsy. While the crucial role of the superior temporal sulcus (STS) in biological motion perception had been documented,³¹ a study that examined the neural correlates of recognition of communicative interactions from point-light motion suggested a significant role of a broad network of cortical and subcortical structures associated with both mentalizing and mirroring networks.³² One of the most robust findings of our study is the deficient classification of individual actions of two agents. Similar effects had been previously found in another clinical population (patients with schizophrenia²³). Interestingly, previous studies in HCs have found that communicative interactions facilitate the detection of a second agent in the CID-5 task,³³ which can be attributed to congruency between actions of both agents.³⁴ Thus, the interpretation of the results from the CID-5 task may be two-fold. On the one hand, it may be seen in terms of overinterpretation of individual actions of agents as communicative ones. However, it has been suggested that the preferential processing of communicative interactions may stem from mechanisms that are linked to the coupling between frontal action observation network and the STS,³⁵ which are rarely affected by MTLE. This way, patients with MTLE may still be able to use communicative gestures of one agent and the congruency between actions of both agents to process vignettes presenting communicative interactions more effectively than individual actions of two agents which carry no additional information that would facilitate processing of the actions of the agents. This interpretation is also supported by our recent findings of dissociation between intact gaze processing (which is a function of the STS) and abnormal emotion recognition (which is linked to the medial temporal lobe structures) in patients with temporal lobe epilepsy (TLE).⁹

The FHA task that we carried out to examine social perception in patients has been previously used by other researchers.⁶ However, in the previous study, FHA were scored qualitatively based on patients’ verbal descriptions of shapes’ actions.³⁶ Here, we used quantitative scoring methods introduced by White et al.,²⁸ which facilitate standardized examination of social perception, compared with scoring based on patients’ verbal descriptions. Despite different scoring methodologies, Broicher et al.⁶ found that patients with MTLE attributed less intentionality to triangle actions than HCs during goal-directed and mentalizing animations, while both groups did not differ in their descriptions of random motion items. In line with these findings, we observed that while the total score for FHA classification task differentiates both groups of MTLE patients from HCs, between-group differences may be found for goal-directed and mentalizing but not random motion animations. Importantly, all of these effects were still observed after controlling for education as a covariate. A comparable pattern of findings was also reported by Bujarski et al.,⁵ who used a video vignette-based task (TASIT); they found that patients with TLE were able to correctly process and comprehend simple sincere social exchanges but were impaired when theory of mind abilities had to be used to correctly process sarcasm or insincere exchanges. Furthermore, we observed a pattern of robust correlations between social cognitive tasks and demographic, cognitive, and clinical factors. CID-5 recognition of interactions was negatively correlated with a mean number of seizures per month and psychomotor speed. The latter result, which suggests a significant role of general visuomotor abilities in inferring the communicative interactions in a clinical sample, is in line with previous research that revealed a similar relationship in a group of patients with schizophrenia.²⁸ Importantly, selection of correct descriptions of specific actions of each agent in the CID-5 task was linked to demographic (years of education) and cognitive (Trail Making Test, Part B time, Digit Symbol score) but not clinical variables, thereby suggesting the role of psychomotor speed and attentional-executive factors for social inference processes in MTLE. For the FHA task, a relationship between demographic (years of education) as well as clinical variables (duration of epilepsy) and FHA categorization was observed. Moreover, a relationship between cognitive (Trail Making Test, Part B) factors and FHA emotion recognition was observed. This finding is congruent with some others⁶ reporting a significant association between Full Scale IQ, years of education, and duration of epilepsy treatment in a group of 42 patients with focal epilepsy.

These results point to a crucial role of nonsocial cognition (i.e., psychomotor speed and executive functions) in social inference in MTLE, especially when precise differentiation of specific actions or emotions is required. This notion is corroborated by the observation that each of the cue-based scores (CID-5 alternative selection, FHA emotion recognition) was significantly associated with cognitive factors. The contribution of both cognitive and clinical factors to social cognitive deficits was previously supported by two tasks that used TASIT to examine social perception in epilepsy. Bujarski et al.⁵ found correlations between TASIT basic emotion interference and complex (but not simple) mental state interference and Full Scale and Verbal IQ in a group of 42 patients with epilepsy (including 32 patients with TLE). A similar pattern of correlations was also found for age of onset of epilepsy. Using the same set of video vignettes, in a group of 87 individuals with temporal lobe epilepsy before or after lobectomy, Cohn et al.²¹ found a significant association between (a) full scale IQ and emotion recognition and (b) comprehension of deceitful and sarcastic exchanges. Moreover, both age of onset and epilepsy duration were found to be linked to each of the scores. The issue of the unique contribution of demographic and clinical variables and medial temporal lobe sclerosis (MTS) to social interference deficits in TLE was also examined by using the voxel-based morphometry method. Both MTS and IQ contributed to deficits in deceit and sarcasm comprehension.

We found that MTLE patients with and without anterior temporal lobectomy present similar deficits in social cognition skills. This type of surgical treatment is performed exclusively in patients with drug resistant epilepsy; therefore, most of the MTLE patients—including our subjects—have been experiencing symptoms of epilepsy for many years, some of them since birth or early childhood. Such a long time of uncontrolled seizures may cause alterations to brain tissue and consequently its functions. In this clinical group, social cognitive skills mediated by circuitry containing affected temporal structures may not have developed properly, so temporal resection probably could not significantly worsen patients’ performance.

Given a risk of various cognitive deficits (language, memory, etc.)^37,38 following temporal resections, our findings provide important information concerning prognosis of postoperative social cognition skills, which can be valuable for physicians and patients considering surgical treatment. One of the main limitations of this study is the small number of participants, which makes our conclusions hard to generalize to the whole population of patients with MTLE. Another issue is a lack of an extra TLE epilepsy control group. It would be interesting to assess whether the social cognition deficits are site specific or present similarities across various epilepsy locations. However, patients were recruited for the study from the population treated at the Department of Neurosurgery, to which only drug-resistant cases are referred. As has been documented, the majority of drug-resistant cases are MTLE.³⁹ Also, in the postoperative group, we have analyzed only patients who underwent ATL. We believe that comparing different methods of treatment, including selective amygdalohippocampectomy or temporal ablation (as they are more restricted than ATL), would provide valuable information for clinicians choosing the method of surgical treatment. Moreover, given that we did not evaluate depressive symptoms in our participants, it would be interesting to investigate the impact of mood states on social cognition in MTLE patients. Future studies could elucidate how depressive symptoms could bias performance on the tasks, especially given the high rates of depression in TLE. Furthermore, due to the brevity of the assessment, we were unable to include methods of functional outcome or quality of life assessment in the battery. Thus, the association between social perception deficits and patients’ functioning should be examined in future studies. Another limitation is that pre- and postsurgical groups were composed of separate sets of patients. It would be very interesting to perform a longitudinal study and repeat assessments in patients who were previously evaluated with the same battery.

Conclusions

We have shown that social perception deficits involving inference from motion cues are a robust finding in patients with MTLE. Deficits were found irrespective of the lobectomy and were still observed after controlling for demographic factors. As has been suggested, both mesial temporal and prefrontal dysfunctions and early interference with cognitive development may underlie social cognitive deficits in epilepsy.⁴⁰ Previous studies documented that social cognitive dysfunction is a stronger predictor of social functioning problems in patients with epilepsy than the general cognitive measures are.⁴¹ Thus, the role of social perception deficit and its association with symptomatology and functional outcome in MTLE should be further studied.