How about watching others? Observation of error-related feedback by others in autism spectrum disorders

Highlights

• We examined the oFRN in participants with ASD and controls.

• Both groups displayed robust oFRN activity.

• oFRN amplitude did not differentiate groups.

• oFRN amplitudes were not correlated with age, anxiety, or autism symptom severity.

• The social and motivational contexts did not limit feedback processing in ASD.

Abstract

Research indicates that individuals with autism spectrum disorders (ASD) may have a reduced ability to utilize performance feedback to regulate their behavior; however, it is unclear to what degree alterations in the environmental context affect feedback processing and contribute to the symptoms of ASD. We utilized the observational FRN (oFRN), an event-related potential (ERP) component that putatively indexes feedback processing while observing feedback directed toward another person, to examine the influence of motivational and social demands on feedback processing in ASD. High-density electroencephalogram recordings were collected from 38 youth with ASD and 31 control participants similar on age and IQ while they observed a confederate performing a modified Eriksen Flanker task. Participants were instructed to count the confederate's errors and were told that they would be awarded based on performance: the confederate would either earn points for the participant or herself. Both groups showed robust oFRN activity on traditional scalp-electrode waveforms and waveforms identified using temporospatial principal components analysis. Amplitude of oFRN did not differentiate groups. Results remained non-significant when comparing medicated to non-medicated participants. There were no significant correlations between oFRN amplitudes, autism symptom severity, and anxiety symptoms. Findings suggest that the social context of the task and motivational significance of the confederate's performance did not limit feedback processing in ASD. Future research in which the context is manipulated further is warranted to determine whether increased environmental complexity influences feedback processing in ASD.

Introduction

A growing body of neurobiological evidence suggests that functional and structural abnormalities in the anterior cingulate cortex (ACC) are prevalent in individuals with autism spectrum disorders (ASD), including disruptions in connectivity (Assaf et al., 2010), metabolism (Levitt et al., 2003), and abnormal activation during a range of cognitive tasks (Ashwin et al., 2007, Dichter and Belger, 2007, Thakkar et al., 2008). The ACC and pre-frontal medial cortex (PFMC) have been tied to cognitive control processes involved in flexibly regulating behavior to achieve goals and adapt to changing environmental demands (Botvinick et al., 2001, Botvinick et al., 2004). These cognitive control processes may involve integrating proprioceptive information regarding self-action, with exteroceptive information regarding the consequences of behavior (Mundy, 2003). Thus, a key component of behavioral regulation during cognitive control involves responding to environmental feedback that signals whether or not behaviors result in a desired outcome. Accordingly, the context, motivational value, and salience of feedback are highly influential in behavioral modification. Recent research indicates that abnormalities in the ACC and PFMC may be tied to impairments in these cognitive control processes in individuals with ASD, contributing to symptomatic difficulty with behavioral flexibility and social–emotional reciprocity.

Studies using electrophysiological measures have identified several indicators of ACC activity associated with feedback processing, including the feedback-related negativity (FRN). The FRN is a negative deflection of the event-related potential (ERP) that peaks approximately 250 ms to 300 ms following stimulus onset and is more pronounced for negative or unexpected outcomes relative to favorable performance feedback (Hajcak et al., 2007, Holroyd et al., 2009, Thoma and Bellebaum, 2012). The FRN tends to be maximal at frontocentral scalp-electrode sites and reflects neural activity in the mesencephalic dopamine system, with the ACC as the primary neuronal generator (Cohen et al., 2007, Gehring and Willoughby, 2002, Nieuwenhuis et al., 2005). FRN amplitudes appear to reflect evaluative processing that distinguishes between good and bad outcomes but not the relative magnitude of outcomes (Hajcak et al., 2006, Hajcak et al., 2007, von Borries et al., 2013). However, several studies indicate that FRN amplitudes are sensitive to motivational, social, behavioral, and emotional factors (Gehring and Willoughby, 2002, Gu et al., 2010, Teper and Inzlicht, 2014, Yang et al., 2013). These contextual factors may affect the motivational impact of reward outcomes, modulating the degree of outcome expectation or significance of good vs. bad outcomes, reflected in changes in FRN amplitude (Yang et al., 2013). Thus, subtle changes in social, motivational, and emotional contexts may affect feedback processing as represented by the FRN (Moser and Simons, 2009).

Three studies have been published to date examining ERP components relating to feedback processing in individuals with ASD (Groen et al., 2008, Larson et al., 2011, McPartland et al., 2012). Groen et al. (2008) evaluated early ERP components associated with feedback in children diagnosed with subthreshold ASD, attention-deficit hyperactivity disorder (ADHD), and typically-developing (TD) controls during a feedback-learning task providing positive and negative feedback. Individuals with ASD differed from TD controls as a function of feedback valence, displaying atypical (i.e., decreased amplitude) ERP response associated with reward anticipation. However, it is difficult to generalize the results of the study due to several limitations including the implementation of a paradigm that failed to elicit a typical FRN response in the TD group and the omission of standard diagnostic procedures. Larson et al. (2011) utilized a guessing task with monetary loss/gain feedback and showed a robust FRN to loss trials relative to gain trials; however, the ASD and TD groups did not differ in FRN amplitude to gain or loss trials. McPartland et al. (2012) also demonstrated a robust FRN to suboptimal outcomes that is similar between individuals with ASD and TD controls regardless of valence.

In contrast, studies of internally generated performance monitoring using the error-related negativity (ERN) demonstrate abnormal performance monitoring in ASD. The ERN is a negative deflection in the ERP that occurs within 100 ms of making an error (Falkenstein et al., 1991, Gehring et al., 1993). Whereas the FRN reflects response to explicit feedback, the ERN relies on early internal monitoring of errors. Several studies have demonstrated decreases in the amplitude of the ERN in individuals with ASD relative to TD controls (Sokhadze et al., 2010, South et al., 2010, Vlamings et al., 2008). Individuals with ASD tend to display less ERP differentiation between error and correct trials than their control counterparts, exhibiting reduced amplitude ERN on error trials but no amplitude differences relative to controls on correct trials (e.g., Santesso et al., 2011, South et al., 2010). Together, these decreases in ERN amplitude and atypical ERP differentiation between correct and incorrect trials could indicate a general deficit in self-monitoring in individuals with ASD. Notably, Henderson et al. (2006) found that ASD children with fewer parent-reported symptoms of social cognitive impairment displayed enhanced ERN amplitudes relative to ASD children with higher reported severity of symptoms regarding atypical social behavior (Henderson et al., 2006). These findings are consistent with the notion that altered functioning in the frontal-cortical networks may lead to concomitant deficits in response monitoring, executive function, and social function (Courchesne and Pierce, 2005, Hill, 2004).

Together studies of the ERN and FRN suggest that individuals with ASD exhibit typical feedback processing but have difficulties with self-related processing or internal regulation of performance. Studies conducted to date, however, do not fully replicate the complex social, motivational, emotional, and behavioral demands in day-to-day situations. Thus, it is yet unclear whether subtle alterations in the social or motivational context affect feedback processing in ASD. Given that individuals display impaired social cognitive functioning, including deficits in social reward processing (Dawson et al., 2005), it is possible that adding requirements to monitor another's performance will disrupt the ability to monitor even concrete, externally generated feedback.

Research investigating the role of the FRN in typical development, using both active and observational learning tasks, have demonstrated that the FRN is not only elicited following feedback regarding individual error, but an observational FRN (oFRN) is also elicited after observing feedback directed toward another person (Kang et al., 2010, Koban et al., 2012, Yu and Zhou, 2006). The oFRN elicited during observation of others' performance is slightly attenuated relative to the FRN generated during an individual's own performance; however, the overall pattern of the ERPs display relatively similar FRN amplitudes during observation of self-referential versus other feedback (Fukushima and Hiraki, 2009, Itagaki and Katayama, 2008, Yu and Zhou, 2006). Larger oFRN amplitudes were observed during a competitive social task when the participant gained a reward or when their competitor lost (Yamada et al., 2011), suggesting that oFRN amplitudes reflect the motivational salience of rewards during competitive social tasks. Interestingly, de Bruijn et al. (2009) found increased activation in the posterior medial frontal cortex independent of the reward outcome (e.g., gain or loss) or whether the error was observed or self-performed, contrasting the motivational salience interpretation of the oFRN (de Bruijn et al., 2009). However, posterior medial frontal cortex activation did increase when individuals made errors that affected another individual relative to errors that affected themselves (Radke et al., 2011). Thus, when witnessing feedback directed toward another individual the observer may rely on additional cognitive and neural processes (e.g., empathy, mentalizing) to infer the emotional state of the individual in response to feedback along with the implications for one's own performance or feedback outcomes (Thoma and Bellebaum, 2012).

The social and motivational demands of observing another add an additional layer to our previous research showing intact FRN in participants diagnosed with an ASD (Larson et al., 2011), as it is possible that the social and cognitive deficits in ASD will reduce their monitoring of even concrete, externally generated feedback within a social context. This is an important first step in understanding feedback processing in ASD in more externally valid situations. Only one study to date has explored this hypothesis. Bellebaum et al. (in press) examined the FRN in a small sample of adults with mild ASDs during an active and observational probabilistic learning task. They observed attenuated FRN amplitudes among individuals with ASD regardless of the social context, pointing to overall reduced feedback processing in ASD. However, their sample was small (n = 10) and may not be representative of the ASD population as a whole, as it consisted of nine women and one man. Accordingly, replication is necessary to determine the influence of the social and motivational context on feedback processing in ASD.

We aimed to extend previous research to better understand the processing of feedback in individuals with ASD in the context of the performance of another individual. During our task participants observed a confederate as they completed a modified Eriksen Flanker task. Previous findings outside of a social context have documented an intact FRN in individuals with ASD (Larson et al., 2011, McPartland et al., 2012) and an attenuated FRN in social and nonsocial contexts (Bellebaum et al., in press). Based on these results, we hypothesized that social cognitive deficits in the ASD group would decrease motivated attention to the confederate and thus lead to attenuated oFRN amplitudes in the ASD group relative to controls.

In addition, in order to determine the motivational value of reward/loss feedback based on the performance of another individual, half of the participants were also required to differentiate between self-referential and other-referential feedback. For half of the participants the confederate's performance resulted in losses or gains for the observer, while for the other half of the participants the confederate's performance was counted for their own benefit and did not influence the observer. We hypothesized that children and adolescents with ASD would display an attenuated oFRN relative to TD controls, suggesting that increased complexity caused by motivational and social demands may result in reduced cognitive control in ASD. We hypothesized that for both groups, points that counted for the observer would generate increased oFRN amplitude relative to points that counted only for the confederate, but that due to decreased perspective taking in ASD, the magnitude of this difference would be less (i.e., an interaction of diagnostic group × point condition).