Abstract
Successful communication requires that conversational partners attend to and use each other’s perspectives. Indeed, conversation requires more negotiation with “other minds” than any other social activity (Malle & Hodges, 2005). Interlocutors must track which information is mutually shared and which information is privileged (i.e., known only to themselves) and modify their behavior accordingly (Clark, 1992). This recognition of shared versus privileged knowledge is critical for efficiently comprehending language, which, on the surface is ambiguous. For example, if a colleague asked me if I wanted to go to “that restaurant” I could be momentarily confused about the restaurant to which she was referring. However, several cues could guide my understanding of my colleague’s intentions. For example, I could refer back to a previous conversation with this person about a restaurant we both wanted to try. Alternatively, in the absence of such a conversation, I could assume she was asking about the Chinese restaurant whose take-out menu was on my desk, clearly visible to both of us, and not the Italian restaurant I just finished looking up online. Thus, although the statement, “that restaurant” is ambiguous, mutually shared information, determined by perspective-taking, can be used to interpret the reference. It is this basic process of thinking about and using a conversational partner’s perspective that allows for communication to be effective. Communicative perspective-taking refers to this ability to actively use information about another’s mental state to interpret language (Nilsen & Graham, 2009) and is distinguished from general perspective-taking, which refers to appreciating another’s mental state without necessarily having to do something with that information.
Despite the fact that developmental research shows that by their 1st year of life, children’s gestures reflect sensitivity to a conversational partner’s perspective (Liszkowski, Carpenter, & Tomasello, 2008) and the seeming ease with which adults interpret language, communicative perspective-taking is not an effortless process (Apperly, Back, Samson, & France, 2008). Rather, the act of using a conversational partner’s perspective is one that requires significant cognitive and attention resources (Lin, Keysar, & Epley, 2010). Accordingly, even though adults are competent social navigators, they show notably egocentric language processing by not restricting themselves to considering only mutually shared information. For example, during comprehension, they will initially consider information that is privileged before later adjusting to a speaker’s perspective (e.g., Keysar, Barr, Balin, & Brauner, 2000; Keysar, Lin, & Barr, 2003). The demand on cognitive and attention resources associated with holding onto and using someone else’s mental state information during an exchange (e.g., Apperly et al., 2008; Lin et al., 2010) has implications for populations whose attention and cognitive resources are weaker, such as individuals with ADHD. Specifically, this finding suggests these individuals will encounter difficulty during communicative exchanges where perspective-taking is required.
Adults with ADHD demonstrate weaknesses in many skills that have been found to underlie successful communicative perspective-taking. Specifically, good attention and executive functioning facilitate use of another’s mental state during conversational exchanges. For example, adults with stronger inhibitory control skills are better able to use the perspective of a speaker (Brown-Schmidt, 2009; see Nilsen & Graham, 2009, for similar results with children). Similarly, individuals with better working memory capacities are more likely to use their conversational partner’s perspective (Lin et al., 2010). When the cognitive demands of communicative tasks are increased, individuals show more egocentric communicative behavior (e.g., Horton & Keysar, 1996; Roßnagel, 2000) and when secondary tasks are used that demand attention resources (e.g., Lin et al., 2010) or working memory (e.g., McKinnon & Moscovitch, 2007), individuals’ ability to reason about and use another’s perspective are compromised. Because adults with ADHD show weaker performance in areas of executive functioning such as inhibitory control and working memory (Biederman et al., 2007; Boonstra, Kooij, Oosterlaan, Sergeant, & Buitelaar, 2010; Nigg et al., 2005), they may not be able to handle the demands associated with communicative perspective-taking and default to more egocentric (i.e., stemming from their own perspective) communication styles.
Understanding the relationship between ADHD symptoms and communicative perspective-taking may shed light on the mechanisms behind interpersonal difficulties found in ADHD. That is, adults with ADHD report less social competency than their peers (Barkley, Fischer, Smallish, & Fletcher, 2006; Friedman et al., 2003) and undiagnosed adults with clinical levels of ADHD symptoms report psychosocial impairment (Able, Johnston, Adler, & Swindle, 2007). As a largely under-researched line of inquiry, it is unclear what social cognitive mechanisms may contribute to the interpersonal difficulties that adults with ADHD face (Uekermann et al., 2010; one exception is the finding that adults with ADHD have difficulty with affect recognition, Rapport, Friedman, Tzelepis, & Van Voorhis, 2002). Within the child ADHD literature, ADHD symptoms are associated with a number of difficulties in social contexts, such as producing off-topic statements and difficulties in maintaining conversational rapport (Bishop & Baird, 2001; Geurts et al., 2004; Mikami, Huang-Pollock, Pfiffner, McBurnett, & Hangai, 2007). In addition, the communicative performance of children with ADHD suggests that they are less able to attend to the perspective of their conversational partners (Purvis & Tannock, 1997; Stroes, Alberts, & Van der Meere, 2003).
The goal of the present study was to examine whether individuals who reported high levels of ADHD symptoms demonstrate perspective-taking difficulties in a communication task and whether their reported symptom levels relate to their demonstrated levels of egocentric communicative behaviors. To address this aim, an interactive communication task was used wherein participants were asked to move objects around a display case based on instructions from a “director” (confederate) who was attempting to make the display look like a picture (Keysar et al., 2000). The display contained some objects that were visible to the director (e.g., a medium-sized bear and a large bear) as well as several objects hidden from the director’s view (e.g., a small bear). Thus, on critical trials where the object description best matched a hidden object (e.g., “small bear”), the participants were required to use the perspective of the director to correctly identify the intended referent (e.g., the medium-sized bear). Failure to do so would suggest difficulties integrating the director’s perspective.
Behavioral measures (objects chosen) and eye movements were recorded. Eye movements provide a sensitive, yet unobtrusive method for tracking information processing and allow us to understand the evaluative process participants engage in when processing language (see Tanenhaus, Magnuson, Dahan, & Chambers, 2000). Thus, we were able to capture transitory information about participants’ orientation of attention as well as final behavioral outcomes.
Method
Participants
A total of 81 undergraduate students participated in this study for course credit or for pay (55 females, 26 males, Mage = 19.75 years, SDage = 2.09). 1
Materials
Participants took part in a communication task with a condirector [(based on task developed by Keysar and colleagues (e.g., Keysar et al., 2000; Keysar et al., 2003)]. A confederate was used so that the instructions given to the participants would be consistent. To ensure that the participants performed as naturally as possible, they were told that the director was another participating student. The participants sat across a table from the director with a 5 × 5 grid of boxes between them (see Figure 1). The grid contained eight objects, all of which were visible to the participants. Two of these eight objects were occluded so that the director could not see them. There were five different constellations of objects that comprised the five experimental sets. Each set involved three instructions to move objects. Two of these instructions were noncritical trials wherein the director asked the participant to move mutually observable objects (thus not requiring perspective-taking). One trial in each set was a critical trial in which the participants were required to use the director’s perspective to correctly determine the referent object. For example, when asked to move the “small bear,” the smallest bear was blocked from the director’s view. On consideration of the director’s perspective, the participant should choose the medium bear because this would be the small bear from the director’s point of view.

Example of the display for the communication task.
An Applied Science Laboratories (ASL; Bedford, MA) Mobile Eye head-mounted eye tracker was used to examine the participants’ eye movements during the communication task. Participants wore goggles that had a small camera lens facing the right eye and another small camera lens facing outwards. A microphone mounted to the goggles recorded sound, synchronized to the video recordings. The two video feeds are interleaved on digital tapes by recording alternating frames, leading to a functional recording speed of 30 frames per second. The direction of eye gaze is calibrated using ASL EyeVision software, which creates a digital video file with a visual overlay of the participant’s focal point and pupil diameter throughout the recording.
The Conners’ Adult ADHD Rating Scales–Self-Report, Long Version(CAARS-S:L; Conners, Erhardt, & Sparrow, 1999) was used to assess the participants’ self-reported ADHD symptoms. This measure contains 66 items organized into several scales, including three that assess Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association, 1994) symptoms of ADHD (DSM-IV Inattentive Symptoms, DSM-IV Hyperactive-Impulsive Symptoms, and DSM-IV Total ADHD Symptoms) and an ADHD Index. Participants rated “how much or how frequently” the various statements (e.g., “I have trouble keeping my attention focused when working”) describe themselves using a 4-point Likert-type scale (0 = not at all, never to 3 = very much, very frequently). Psychometric properties of the CAARS-S:L are strong (Kooij et al., 2008).
Procedure
Participants were told that they would be taking part in a communication task in which the director would try to make the final grid display look like a picture by asking the addressee (i.e., the participant) to move objects. The photographs reflected the perspective of the director, clearly depicting the doors occluding some boxes. To highlight the director’s visual perspective, the participant and the confederate each played the role of director for two practice trials. During the practice trial, the confederate pretended to be unsure of the name of an object (saying “What is this called?”) to help create the illusion of naivety. Following practice trials, the participant and confederate chose roles, by a fixed drawing that made it appear the role assignment was random. Participants were then fitted with the eye tracker, and calibration was completed. This portion of the video was later used with the EyeVision software to determine the participant’s focal point during the task. The director was told, in front of the participant, that she would not be allowed to provide clarification to discourage participants from asking questions directly related to perspective.
After the calibration procedure, the confederate was blindfolded and the experimenter placed the first set of objects into the display case. Once all of the objects were positioned, the confederate removed her blindfold and gave directions to the participant. The requests made by the confederate were scripted and in a fixed order. After the three instructed moves, the confederate replaced her blindfold and the procedure was repeated for the remaining four object configurations.
Once the communication task was completed, the confederate left the room, ostensibly to perform the final tasks independently from the participant. The participant then filled out the CAARS-S:L, after which he or she was debriefed on the true nature of the confederate’s role.
Coding
The temporal window of observation was the beginning of the noun that described the target object (e.g., the “b” in “bear”) to the point at which the object was lifted up. Two research assistants, blind to the participants’ reported ADHD symptoms, coded video recordings. Adobe Premiere was used as it allows for video–audio synchrony and frame-by-frame analysis. Intraclass correlation (ICC) coefficients were used to establish the level of agreement between the two coders. ICC coefficients (based on 13 participants, 21% of calibrated participants) ranged from .78 to .99.
Results
All participants were assessed for their ADHD symptoms at the time of testing using the CAARS-S:L. Of the 81 participants who participated in the study, 12 were excluded as their CAARS-S:L DSM-IV ADHD symptom scores during testing fell within the borderline range (T-score of 60-64). The resulting sample was 30 students who reported ADHD symptoms that fell within the clinical range, T-scores > 65, forming the “high-ADHD-symptom group”; 20 females, 10 males, Mage = 19.63 years, SDage = 2.48; MT-score = 77.22, SDT-score = 8.36, and 39 students with self-reported ADHD symptoms within the average range, T-scores < 60, forming the “low-ADHD-symptom group”; 28 females, 11 males, Mage = 20.00 years, SDage = 1.82; MT-score = 49.49, SDT-score = 6.71; t(67) = 15.30, p < .001. 2 Six participants in the high-ADHD-symptom group indicated that they had been previously diagnosed with ADHD. The T-scores of the two groups also differed significantly on the DSM-IV Inattentive Symptoms index, T-score difference of 24.3, t(69) = 10.92, p < .001, and on the DSM-IV Hyperactive-Impulsive Symptoms index, T-score difference of 23.3, t(43) = 11.86, p < .001. Eye movement data from 12 participants from the high-ADHD-symptom group, and 8 participants from the low-ADHD-symptom group could not be used because the eye tracker was unable to accurately track their pupils, consistent with calibration rates of previous eye tracking work (e.g., Lin et al., 2010). Scores greater than three standard deviations from the mean were removed as outliers.
The variables of interest in the communication task were as follows: (a) the number of fixations on the target or blocked object, (b) the response latencies to select the target object, and (c) the objects chosen. Eye movement variables were calculated only from the trials in which participants correctly chose the target object. Thus, our variables capture participants’ online attention to the objects as well as their behavioral accuracy. We compared participants’ performance on critical trials (where perspective-taking was required) with their performance on noncritical trials (where the best referent for the direction was a mutually observable object). We expected that critical trials would lead to more interpretative errors (e.g., considering, or choosing, blocked objects) due to the added cognitive burden of having to take into account another’s perspective. Furthermore, we anticipated that these trials would be particularly difficult for individuals who reported high ADHD symptoms.
Number of Fixations
A 2 (group: high/low ADHD symptoms) × 2 (object: target/blocked) × 2 (condition: critical/noncritical) mixed ANOVA revealed a main effect of object, F(1, 44) = 290.76, η2 P = .87, p < .001; condition, F(1, 44) = 28.20, η2 P = .39, p < .001; and group, F(1, 44) = 7.23, η2 P = .14, p < .01 (see Figure 2). The main effects were qualified by significant interactions between object and condition, F(1, 44) = 6.90, η2 P = .14, p < .05, and between condition and group, F(1, 44) = 5.48, η2 P = .11, p < .05. The first interaction was followed up with paired-sample t tests across groups with the Bonferroni correction. Participants fixated on the targets more than blocked objects in critical, t(45) = 11.41, d = 1.84, p < .001, and noncritical conditions, t(48) = 20.50, d = 3.34, p < .001. Blocked objects were fixated on more during critical trials than noncritical trials, t(45) = 5.34, d = 0.99, p < .001, but this effect of condition did not reach significance when looking at fixations toward the target (p = .06). The interaction between condition and group was followed with independent-sample t tests (with Bonferroni correction) between the groups collapsed across object. In the critical condition, participants with high ADHD symptoms produced more fixations than participants with low ADHD symptoms, t(44) = 3.16, d = 0.92, p < .005. Conversely, on noncritical trials, the groups did not differ, p = .42. Of note, planned comparisons on fixations toward target and blocked objects in the critical condition revealed a significant group difference on the number of fixations toward the blocked object, t(44) = 2.55, d = 0.72, p < .05, but not the target object (p = .09). Thus, adults who reported high ADHD symptoms had less efficiency on trials in which perspective-taking was required and this finding was driven by a greater propensity to fixate to the “egocentric” (i.e., blocked from the speaker) object.

Number of fixations toward target and blocked object in critical and noncritical trials.
Response Latencies
The amount of time it took participants to hone in on the target (i.e., from onset of the noun to the time participants lifted the target) was assessed in a 2 (group) × 2 (condition) mixed ANOVA. Results revealed a main effect of condition, F(1, 61) = 51.00, η2 P = .46, p < .001, whereby participants took longer to choose the target in the critical trials (M = 4,540 ms, SD = 816 ms) than in the noncritical trials (M = 4,002 ms, SD = 602 ms). A significant effect of group was not demonstrated (p = .20) nor was there a significant interaction between the variables, p = .19. Thus, having to take the director’s perspective created an added processing burden for participants overall, but this did not significantly interact with ADHD symptomatology.
Object Choice
Participants’ errors in object choice were assessed across critical and noncritical trials in a 2 (group) × 2 (condition) mixed ANOVA. Results revealed a main effect of condition, F(1, 67) = 25.46, η2 P = .28, p < .001, but neither group, p = .49, nor a significant interaction, p = .54. Here, participants were more likely to make errors on the critical trials (M = 21%, SD = 34%) than on the noncritical trials (M = 0.1%, SD = 0.1%). Thus, in general, participants made more egocentric errors than general interpretive errors.
Relationships Between ADHD Symptoms and “Egocentric” Communicative Performance
Of interest was the degree to which participants’ self-reported ADHD symptoms related to their communicative perspective-taking performance. Recall that increased egocentric processing would be reflected by greater consideration of (i.e., increased fixations toward) an object that was blocked from the director’s view.
A number of CAARS-S:L subscales significantly correlated with the number of fixations made toward the blocked object during critical trials: DSM-IV Inattentive, r(45) = .45, p < .005, DSM-IV Hyperactivity-Impulsivity, r(45) = .30, p < .05, DSM-IV ADHD Symptoms, r(45) = .41, p < .01, and the ADHD Index, r(45) = .45, p < .005. Importantly, all these correlations remained significant when participants’ number of fixations to the target objects on the noncritical trials (i.e., reflecting general processing) was covaried out (all ps < .05).
However, a number of significant correlations between participants’ number of fixations to the target object in the critical trials and their ADHD symptoms were also found, including, DSM-IV Inattentive, r(45) = .35, p < .05, DSM-IV Hyperactive-Impulsive, r(45) = .30, p < .05, DSM-IV ADHD Symptoms, r(45) = .33, p < .05, and ADHD Index, r(45) = .30, p < .05. Thus, individuals with higher ADHD symptoms showed less certainty, as reflected by more frequent fixations toward objects generally. To assess whether the egocentric object creates more of an attention pull, when controlling for uncertainty on critical trials, we correlated participants’ self-reported ADHD symptoms and fixations on the blocked object in the critical condition when controlling for fixations toward the target object in the critical condition. Significant correlations remained between fixations to the blocked object and individuals’ self-reported DSM-IV Inattentive Symptoms, r(43) = .37, p < .05, DSM-IV ADHD Symptoms, r(43) = .32, p < .05, and the ADHD Index, r(43) = .34, p < .05. Thus, when fixations toward the target object in critical trials are controlled for, individuals with greater ADHD symptoms, in particular inattention, show greater fixations toward the egocentric object.
The percentage of times that participants chose the egocentric object was not significantly correlated with any of the self-reported ADHD symptoms (all ps > .56). Thus, participants’ self-reported ADHD symptoms did not relate to behavioral errors made during the communicative task.
Discussion
The hypothesis that individuals with self-reported ADHD symptoms would use a conversational partner’s perspective less efficiently when interpreting statements was supported. Specifically, individuals with high levels of ADHD symptoms looked at an object blocked from the speaker’s view more frequently than did individuals with low levels of ADHD symptoms during an interactive communicative task. Furthermore, with greater reported impairment in attention came more frequent egocentric communication processing. ADHD symptoms, however, did not ultimately affect object choices.
Consistent with previous literature, all participants found it more difficult to interpret statements when perspective-taking was required (e.g., Keysar et al., 2000). That is, participants, regardless of ADHD symptomatology, looked at a blocked object more frequently, took longer to choose the target object, and made more errors, when the trials required attention to the director’s visual perspective. According to the “anchoring and adjustment” heuristic, the use of perspective-taking within conversation does not happen automatically (Epley, Keysar, Van Boven, & Gilovich, 2004). Rather, individuals first assess information from their own perspective and then serially adjust to the perspective of their conversational partner. In this way, individuals continue to attend to privileged information, rather than limiting their attention to mutually shared information, as shown by our data.
Central to our hypothesis, individuals with high levels of self-reported ADHD symptoms showed more fixations on critical trials across object types (hidden or target), whereas differences between groups were not found on noncritical trials. This suggests that it was the ambiguity of the critical trials, rather than general issues with language processing or oculomotor control (which have been found to be disrupted in children with ADHD; Rommelse, Van der Stigchel, & Sergeant, 2008), that led to these group differences. The eye movements of individuals with high ADHD symptoms suggest greater uncertainty when deciding on the intended object. It appears that these participants did not use the director’s perspective efficiently to hone in on the target object, resulting in more fixations on the objects before making their choices. Furthermore, even when controlling for the increased processing in general on critical trials, inattentive symptoms were specifically related to fixations on the blocked object during these trials. That is, the individuals who report higher degrees of inattention fixate more on the egocentric object suggesting that they are less efficient with their use of perspective information within a communicative context. Although the CAARS:S-L contains several items related to communication (e.g., “I blurt out things”) and social functioning (e.g., “I have trouble waiting in line or taking turns with others”), these items are not included in the DSM-IV Inattentive Symptoms subscale, which was the subscale found to be the most correlated with participants’ communicative perspective-taking performance. This strengthens the argument that underlying attention difficulties may be responsible for the communicative difficulties seen in ADHD.
Within interactive contexts, perspective-taking must be expedient to guide interpretations of rapidly changing social contexts, efficient enough so that not all cognitive resources are being utilized and flexible enough to allow for changes in reasoning about another’s behavior (Apperly & Butterfill, 2009). Furthermore, given the essential nature of perspective-taking for social interaction (Mead, 1934), one would expect difficulties in perspective-taking to lead to pro-blematic social behaviors or to reduce socially competent behaviors. For example, difficulties with communicative perspective-taking could mean that misunderstandings are more frequent or that individuals face greater challenges in contexts where perspective-taking is essential, such as interpreting the intentions of a sarcastic speaker, collaborating with individuals who may or may not share goals, or negotiating with others. Indeed, previous work has shown that strategies to increase perspective-taking affect individuals’ cooperative behavior (e.g., Epley, Caruso, & Bazerman, 2006). Furthermore, within the child literature, communicative perspective-taking is related to later social cognitive skills (Bernard & Deleau, 2007) suggesting that challenges in this area may relate to more global social processing difficulties. However, at the initial stage of this research, it remains to be determined how the results from the current study relate to the more general interpersonal difficulties that are reported for people with ADHD. For example, although individuals with high ADHD symptoms showed communicative processing that was more egocentric, we did not observe these individuals to actually choose the blocked objects more often. As such, it will be important for future work to examine how efficiency in communicative perspective-taking relates to other aspects of social interactions.
Several researchers have identified a distinction between having perspective-taking ability versus using perspective information, in that, during communication, individuals with reasonable perspective-taking skills could have difficulty with applying this information if cognitive resources are taxed (e.g., Horton & Keysar, 1996; Lin et al., 2010; Nilsen & Fecica, 2011). Given this distinction, present results do not necessarily suggest that individuals with high levels of ADHD symptoms lack the awareness of others’ perspectives. In fact, it has been shown that children with ADHD do not differ from their peers on theory of mind tasks (Perner, Kain, & Barchfeld, 2002; Sodian & Hülsken, 2005). Thus, it is likely that the ability to use perspective information effectively is what differs for individuals with high levels of inattention and/or hyperactivity, as opposed to their ability to appreciate a conversational partner’s mental state per se. As research has demonstrated that children with ADHD have difficulty in suppressing the desire to look at certain visual stimuli in oculomotor inhibition tasks (Mahone, Mostofsky, Lasker, Zee, & Denckla, 2009), it may be the case that individuals with high ADHD symptoms have difficulty suppressing their own perspective leading to more looks toward the blocked object. Thus, inhibition as well as other areas of cognitive weakness for individuals with ADHD (e.g., working memory; Biederman et al., 2007; Boonstra et al., 2010; Nigg et al., 2005) may account for the relative difficulty in communicative perspective-taking found here in individuals with high levels of reported ADHD symptoms.
This study involved the self-reported symptoms of a nonclinical sample of university students, which affects some conclusions. First, inferences about clinical populations of individuals with ADHD are somewhat limited. Our high-ADHD-symptom group reported symptoms that fell within the clinical range, and research has found that undiagnosed adults with clinically elevated symptoms of ADHD show similar levels of psychosocial impairment as those diagnosed with ADHD (Able et al., 2007). However, given that this study used a university population, it may be that our high-ADHD-symptom group has less functional impairment than clinical populations, as evidenced by their level of academic achievement. As such, we suspect that the results would be more pronounced in clinical populations, with adults with diagnosed ADHD showing higher levels of difficulty using perspective in communication. Second, we did not conduct a full assessment of comorbid psychiatric concerns or neuropsychological characteristics which may relate to the observed findings. However, it is unlikely that comorbid features (such as negative affect) would fully account for the observed results as previous research has found that a negative mood produces less egocentric processing than a positive mood (Converse, Lin, Keysar, & Epley, 2008). With respect to neuropsychological factors, it may be that differences in neuropsychological profiles of our groups relate to their different communicative perspective-taking performance (e.g., see Nilsen & Fecica, 2011). Indeed, we could expect that differences in our groups on measures of executive functioning may mediate the relationship between ADHD symptoms and egocentric communication behavior. Examining the interrelations between these factors is an important avenue for future research.
Conclusion
In sum, successful communication requires that conversational partners appreciate and use each other’s perspective in an online fashion. We have shown that individuals who report clinically significant levels of ADHD symptoms are less efficient with how they integrate a speaker’s perspective into their interpretations, likely due to the significant processing demands associated with this aspect of communication.
Footnotes
Acknowledgements
The authors thank the students who participated in the study. They also thank L. Flannery, T. Le, V. Li, R. Neal, F. Preston, N. Siva, and Z. Semnani-Azad for their assistance with this research.
Authors’ Note
Data from this study were presented at the Canadian Psychological Association 2010 Meeting.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by funding from an Ontario Mental Health Foundation New Investigator Grant awarded to E.N.
