فراشناخت نمایندگی و تئوری ذهن در افراد بزرگسال مبتلا به اوتیسم عملکرد بالا

We investigated metacognition of agency in adults with high functioning autism or Asperger Syndrome (HFA/AS) using a computer task in which participants moved the mouse to get the cursor to touch the downward moving X’s and avoid the O’s. They were then asked to make judgments of performance and judgments of agency. Objective control was either undistorted, or distorted by adding turbulence (i.e., random noise) or a time Lag between the mouse and cursor movements. Participants with HFA/AS used sensorimotor cues available in the turbulence and lag conditions to a lesser extent than control participants in making their judgments of agency. Furthermore, the failure to use these internal diagnostic cues to their own agency was correlated with decrements in a theory of mind task. These findings suggest that a reduced sensitivity to veridical internal cues about the sense of agency is related to mentalizing impairments in autism.

This article addresses the question of whether people with autism spectrum disorders (ASDs) exhibit difficulties in distinguishing between self-controlled and externally-controlled action, that is, whether they may exhibit abnormalities in their metacognition of agency. The ability to reliably monitor whether one is, oneself, controlling an action, or whether the action is being controlled or interfered with by forces or circumstances that are external, seems to be fundamental in establishing self-other boundaries. Difficulties in establishing and maintaining such boundaries between self and non-self could have many consequences, including impairments in understanding the perspective of another and in knowing that the perspective of another is different from one’s own perspective – difficulties that are prominent in people with ASDs. Russell (1996) proposed that an impairment in agency monitoring processes in individuals with ASDs might emerge as early as the second year of life. He suggested, as well, that an agency detection breakdown could impact the acquisition of theory of mind (ToM, i.e., the ability to attribute mental states to oneself and to others) (Premack & Woodruff, 1978) in ASDs, because mentalizing abilities centrally depend upon possessing a form of pre-theoretical self-awareness that self-agency-monitoring makes possible. This ability underpins the most basic processes needed for ToM, and hence deficits in one’s own sense of agency may be deeply linked with ToM impairments (Russell, 1996). This proposal is consistent with what is known about the developmental trajectory of agency monitoring in normal children. Human infants are not initially endowed with either a fully mature ToM, or sense of agency. An elementary sensitivity to one’s own agency, which arises in infants around 9–18 months of age (Johnson, 2003), is thought to be a developmental precursor of the ability to assign intentions to others and to be capable of making metacognitive judgments about one’s own of agency. The experience of being the agent of one’s own action, as compared to the experience of being an observer of an event caused by external sources, is constitutive of the self-other differentiation. Baldwin (1906) argued that for an infant to overcome adualism (the merging of objectivity and subjectivity), he or she must discover that data from the world are sometimes discrepant with his or her needs (i.e., that the nipple is not invariably there). It is the mismatch between the child’s ‘conative affective striving’ and what is there that allows the child to begin to separate the self from the outside world, and that provides a precursor for the development of a self. Thus, the detection of discrepancy between one’s own intentions and what happens (due to others or the world) allows the individual to begin to sort out the relation between the self and others as is needed for a formation of self-awareness at the most elementary level. Such discrepancy detection is at the core of a number of models concerned with people’s sense of agency (described below). Any impairment in making a clear distinction between the self and others – i.e., partial perseveration of adualism – could result in (a) conflation of one’s own perspective with that of the other including misattributions of agency, and in (b) difficulties in tasks, such as ToM tasks, in which it is necessary to appreciate that the other’s beliefs or perspectives are different from one’s own. Pacherie (1997), too, has argued that agency monitoring and mentalizing deficits may be impaired, in tandem, in ASD. In her view, it is a specific aspect of agency monitoring that potentially gives rise to elementary self-knowledge and that is related to both metacognition of agency and mentalizing. It is not just the individual’s prior intentions, per se, that are important, but rather their ‘intention in action’ (Searle, 1983) that allows the development of self-awareness. The ‘intention in action’ cue is the ongoing detection of potential discrepancy or lack of discrepancy between the outcome and the intention which indicates the effectiveness of the self’s motor action during action performance and that conjoins self-awareness, agency judgments, and ToM. It is this cue that we tap – as the sensory motor discrepancy detection cue – in the present experiment. Although the rationale for expecting a correspondence between an impaired sense of agency and diminished ToM in ASDs is clear, the empirical studies directed at this question – particularly at the agency question – are few and the results are mixed. Considerable research points to a deficit in ToM among people with ASDs, including those with high functioning autism and Asperger Syndrome (HFA/AS). ASDs are neuro-developmental disorders that are characterized by qualitative impairments in communication, social interactions and stereotyped repetitive behavior (DSM-IV, American Psychological Association, 2000). Indeed, many researchers have regarded impairments in TOM, or in what is sometimes been called ‘mentalizing’, as the core deficit underlying ASDs (Baron-Cohen, 1989, Baron-Cohen, 1995, Baron-Cohen et al., 1985, Frith, 1989, Leslie, 1987, Leslie, 1991 and Leslie and Roth, 1993). Adults with HFA/AS can pass first- and second-order mindreading tests (Dahlgren & Trillingsgaard, 1996), but they often may fail in more ‘advanced’ ToM tasks, based on the detection of sarcasm, irony or bluff (Happé, 1994). They also often fail to appreciate inappropriate or insensitive social comments, that is they do not understand Faux Pas (Baron-Cohen et al., 1999 and Zalla et al., 2009). The Faux Pas advanced ToM task will be used in the present study. While a deficit in ToM is well established in studies of ASDs, the sense of agency has been less investigated. In the present study, we will first overview current ideas on how people are thought to be able to determine whether they are the agent controlling a phenomenon, and then turn to studies investigating this capability in participants with ASDs. The predominant theory explaining people’s sense of agency is the Comparator Model (Frith, 2005, Frith et al., 2000 and Wolpert et al., 1995) of motion control. One important component of this model is that people use an efference copy that incorporates an intention or plan of the action to predict the sensory consequences of a given motor command. This expectation or plan is compared, in real time, to the so-called afference, that is, the actual sensory-motor outcome ( Wolpert et al., 1995). The matching process between central motor plans and the multisensory feedback signals (visual, tactile and proprioceptive) arising during action execution, together with the associated motor intention, is the crucial mechanism thought to both underlie the kind of delicate motor control exhibited by humans and a signal that is the basis for people’s sense of agency. This matching process is tantamount to what Searle (1983) calls ‘intention in action’. Recent theories directed at people’s metacognition of agency offer support for the idea that people use a variety of cues to make conscious judgments of agency. One of these cues could be the match/mismatch cue from the comparator model. Such a cue allows control of motor or other behavior and it can affect behavior at a level prior to awareness (Fourneret and Jeannerod, 1998, Georgieff and Jeannerod, 1998, Jeannerod, 1999 and Sebanz et al., 2006). However, the high-level metacognitive judgment process by which people are able to assess the extent of their control is thought to involve consciousness (Metcalfe, 2013, Metcalfe et al., 2013 and Synofzik et al., 2008), and to be reportable by the individual. Thus, there are two distinct components to agency: (1) an implicit ‘feeling of agency’ (FoA), mediated by lower-level, pre-reflective, sensori-motor processes that subserves action-processing, and which arises mainly from the match or mismatch between the efference copy and afference, and (2) an explicit or conscious ‘judgment of agency’, which is based, not only on a signal from this matching process but also on reflective or belief-like processes. These explicit metacognition of agency judgments can use but are not limited to using only the match/mismatch cue. Other cues can also contribute to these judgments. Importantly, some of the other cues that are utilized in the conscious judgments of agency are the perceived goodness of performance and the salience of reward (for example, see Metcalfe, 2013). These ‘external’ cues are not necessarily accurate with respect to the self/other distinction. For instance, performance can be accidentally good without the self being implicated in causing the good performance. However, the discrepancy cues described in the comparator model (i.e., cues signalling a discrepancy between efference and afference) do provide the actor with diagnostic information concerning his or her own agency or lack of agency. In the experiment that follows these critical discrepancy cues are captured in the turbulence and lag conditions, which involve distortions in objective control. The individual’s reliance on these diagnostic cues can be contrasted with his or her reliance on cues such as the perceived goodness of performance, which are not necessarily diagnostic of agency. There are some indications – though few are direct or compelling – that there may be impairments in metacognition of agency associated with ASDs. This evidence is often indirect, and no previous studies have sought to distinguish whether the cues that people with ASDs are using are diagnostic or non-diagnostic. Prominent among the experiments providing indirect evidence is a study by Russell and Jarrold (1999) who found that children and adolescents with autism had difficulties remembering whether they or another person had performed certain actions and thus, unlike the comparison groups, were not better at recalling their own actions. Such a failure might have been a selective memory abnormality, but it might also have occurred because the children with ASDs did not adequately monitor and distinguish between what they had done and what the other had done in the first place (see, Haswell, Izawa, Dowell, Mostofsky, & Shadmehr, 2009). Note, though, that Hill and Russell (2002) failed to replicate this study, and that Williams and Happé (2009) also reported typical memory effects in participants with ASDs. Other studies, however, have shown – like the original Russell and Jarrold (1999) study – that ASDs are associated with smaller differences in memory than those shown by typical participants as a function of self-relevance (Hare et al., 2007, Millward et al., 2000 and Toichi et al., 2002). Individuals with typical development remember and apparently give preferential processing to information that is self-generated, self-relevant, or self-enacted, whereas individuals with ASDs fail to accord such preferential treatment and sometimes cannot discriminate self-generated from the other-generated information (Millward et al., 2000 and Toichi et al., 2002). It is possible that the ASD participants exhibit this deficiency because, as Zalla, Daprati, Chaste, Nico, and Leboyer (2010) hypothesized, they have a diminished sensitivity to internal agency cues (i.e., a deficit in metacognition of agency). Such diminished sensitivity would reduce the difference between observed and performed actions in memory. Consistent with this hypothesis, Zalla and collaborators reported that individuals with HFA/AS, when asked to recall self-performed actions, did not profit from performing the actions themselves to the same extent as did typically developed individuals and failed to encode personal events in the same way as a matched control group. However, while this reduced self-reference in ASDs is plausible, alternative interpretations have also been offered, including impaired episodic memory (Bowler, Gardiner, & Grice, 2000), delayed development of source monitoring abilities (Bowler et al., 2004 and Lind and Bowler, 2009) or difficulties in complying with the executive demands of various self-referential tasks (Hala, Rasmussen, & Henderson, 2005). Several other studies have more directly investigated the possibility that ASDs affect people’s sense of agency. David et al. (2008) constructed a task in which participants were asked to distinguish between their own online action toward a target at the top of the screen and a ‘computer’ action toward the same target. This task, which would seem to be useful in directly assessing breakdowns in agency monitoring, showed comparable performance between ASD and control participants. There are, of course, many reasons why null results might have been obtained. More recently, Sperduti, Pieron, Leboyer, and Zalla (2014) found positive evidence for impairment in ASDs on a different ‘sense of agency’ experiment. They tested individuals with ASDs and controls in an intentional binding task. Intentional binding refers to the finding, in individuals with typical development, of a decrease in perceived time between the onset of an action effect to the offset of the effect when one is, oneself, the agent who initiated the effect, as contrasted to when one observes the same events externally (see, Haggard et al., 2002, Haggard and Eimer, 1999 and Moore and Haggard, 2008 for possible explanation of this effect). Whatever its explanation, this decrease in perceived time as a function of being the agent rather than being a mere observer is widely accepted as reflecting one’s implicit sense of agency. In Sperduti et al. (2014) study, the participants with ASDs exhibited reduced intentional binding, suggesting an altered sense of agency. In summary, although some findings, then, suggest that there might be a deficit in the sense of agency in ASDs, the evidence is scanty, and whether there is a relation between deficits in metacognition of agency and impairments in ToM in ASDs is still unknown. The present study investigated whether individuals with ASDs exhibit impaired metacognition of agency. We were especially interested in whether this purported impairment might be due to an inability to use those cues that provide valid information concerning their own control, while the use of goodness of performance cues might be intact. Accordingly, in the present experiment, a group of adults with high functioning autism (HFA) or Asperger Syndrome (AS) and a group of matched control participants were asked to perform a computer-based agency task in which X’s and O’s streamed from the top of the computer screen. Participants moved the computer mouse, which in turn moved a box on a horizontal bar on the screen (see Fig. 1). As in previous studies (Kirkpatrick et al., 2008, Metcalfe et al., 2010, Metcalfe et al., 2013 and Metcalfe et al., 2012), they were instructed to try to touch all of the falling X’s with the box and avoid touching any of the O’s. After engaging in this task for a short period of time they were asked to make judgments of their performance (JOPs) and judgments of their control or agency (JOAs). Full-size image (13 K) Fig. 1. A screen shot of the task. The participant moves the square on the grey bar at the bottom of the screen to ‘catch’ downward scrolling X’s and avoid catching O’s. Figure options One reason that past research may sometimes not have shown an impairment in the sense of agency in ASDs, may have been because previous experiments had not specifically targeted the online discrepancy detection cue. The present task does so: It includes four conditions, two conditions in which turbulence and two conditions in which lag distorts the synchronous match between the intent and the outcome, or between the efferent and afferent copy, as posited by the ‘Comparator Model’. The extent to which participants could objectively control the cursor using the mouse varied by conditions. In the control condition, there was a perfect correspondence (both spatially and temporally) between participants’ movement of the mouse and the movement of the cursor on the screen. In the experimental conditions, participants’ control of the cursor was altered by the introduction of a brief time Lag between the mouse and cursor movement (in conditions Lag 1 and Lag 2) or by the addition of turbulence or random noise to the cursor position (in conditions Turb 1 or Turb 2). Finally, the experiment included a Magic condition, in which X’s disappeared whenever the cursor was near, but not necessarily touching them (thus, artificially inflating participants’ performance). Within each block, four trials per condition were presented in a quasi-random order, with lag conditions always preceding turbulence conditions. This task allows us to investigate people’s sensitivity to the diagnostic discrepancy cues implicated in the model by contrasting the experimental conditions to a control condition in which no such discrepancies are introduced. We hypothesized that people with ASDs might fail to use the discrepancy cues, associated with the turbulence and lag conditions, in making judgments of agency. Even though such judgments occur at a conscious level, the disturbance might arise at the preconscious source of these sensorimotor cues – detection of discrepancy in the comparator model – leading to motor dysfluency and abnormalities across a wide range of behaviors in people with ASDs (Fournier, Hass, Naik, Lodha, & Cauraugh, 2010). Importantly, as noted above, only some of the cues that are integrated into the judgment of agency are thought to be diagnostic, that is to actually allow distinguishing between the self and others (or a computer program) as the cause of the action. Other cues, including perceived goodness of performance, contribute to the conscious judgments of agency, but are not necessarily diagnostic of agency. Regression analyses conducted in past experiments using this paradigm with typical participants have shown that these diagnostic cues contribute to people’s judgments of agency ( Metcalfe et al., 2013). In contrast, patients with schizophrenia – a population known for difficulties in accurately ascertaining their own agency – fail to use these diagnostic cues in making judgments of agency, while, at the same time, they do use non-diagnostic cues such as perceived goodness of performance, to make JOAs ( Metcalfe et al., 2012). The final concern of the present research was with the possible relation, proposed by Russell (1996) and by Pacherie (1997), between abnormalities in the sense of agency and deficits in ToM in ASDs. In the present study, mindreading abilities were evaluated by using an advanced ToM task, the Faux-pas Recognition Task (Baron-Cohen et al., 1999). Previous studies using the Faux Pas recognition task (Baron-Cohen et al., 1999 and Zalla et al., 2009) showed that individuals with high functioning autism have difficulty in reasoning about others’ mental states and emotions, as exemplified by their performance on this test.

The present study reveals that, despite a good degree of metacognition of agency, participants with HFA/AS have a diminished agency monitoring. In making their judgments of agency, they used external cues such as goodness of performance to the same extent as did control participants, but they relied to a lesser extent than did control participants on the particular internal sensory motor cues that are diagnostic of agency. In addition, lower performance in ToM was correlated with reduced diagnostic agency beta scores: the more participants picked up on diagnostic cues to agency – whether the self was indeed in control or not – the better they were at inferring others’ intentions in social situations. These findings support the notion that there is a causal and functional relation between ToM and the sense of agency.