New Frontiers in Neuropsychological Assessment: Assessing Social Perception Using a Standardised Instrument, The Awareness of Social Inference Test
Skye McDonald, School of Psychology, University of New South Wales, Sydney, NSW 2052, Australia. Email: firstname.lastname@example.org
Many clinical disorders lead to impairments in social perception that are not easily detected using standard neuropsychological instruments. In this article, The Awareness of Social Inference Test (TASIT) is described, and evidence for its reliability and validity is reviewed. TASIT was designed to assess the ability to interpret emotional expressions and paralinguistic cues in order to make judgements about speakers' mental states and their conversational meanings, especially when these diverge from the literal meaning of what they say (as in the case of sarcasm). TASIT has three parts: (1) the Emotion Evaluation Test (six basic emotions); (2) Test of Social Inference (Minimal), assessing the ability to understand emotions, mental states, and conversational inferences in sincere and sarcastic interchanges; and (3) Test of Social Inference (Enriched), assessing comprehension of lies and sarcasm, each with alternate forms for re-testing. TASIT has adequate test–retest and alternate forms reliability. Evidence for construct validity is derived from its relationship to conventional neuropsychological tests, experimental tests of social perception, and real-world performance. In addition, research using TASIT to examine deficits in social perception in traumatic brain injury, schizophrenia, frontotemporal dementia, Alzheimer's disease, and stroke attest to its sensitivity as a clinical assessment tool.
Social perception or social cognition refers to the ability to attend to and interpret a range of social cues, including language, paralinguistic information (e.g., intonation), and non-verbal signals (e.g., facial expression, eye gaze, and gesture) (Archer & Akert, 1977; Boice, 1983; Edinger & Patterson, 1983; Morrison & Bellack, 1981; Muehlenhard, Koralewski, Andrews, & Burdick, 1986). Social perception is a cornerstone of good social skills (Boice, 1983; Kagan, 1984; Kahn, 1970; Morrison & Bellack, 1981; Trower, 1980). Without accurate social perception, individuals may respond to others in a manner that is ineffective, poorly timed, or frankly inappropriate. Importantly, specific deficits in social perception have been increasingly reported in a wide range of clinical conditions, including traumatic brain injury (TBI) (Green, Turner, & Thompson, 2004; Ietswaart, Milders, Crawford, Currie, & Scott, 2008; Milders, Fuchs, & Crawford, 2003; Watts & Douglas, 2006), schizophrenia (Brune, 2005a; Edwards, Jackson, & Pattison, 2002), autism and Asperger's syndrome (Baron-Cohen, Jolliffe, Mortimore, & Robertson, 1997), and frontotemporal dementia (FTD) (Gregory et al., 2002), and represent an important target for assessment and remediation.
The problem for clinicians is that while there are some tools of social perception for research use, such as the Mind in the Eyes test (Baron-Cohen et al., 1997) or the Ekman and Friesen photos of facial affect (Ekman & Friesen, 1976; Young, Perret, Calder, Sprengelmeyer, & Ekman, 2002), there is a dearth of assessment instruments that are sensitive to clinical disorders of social perception that have normative data or proven reliability and validity. In important ways, social cognition is qualitatively different to the kinds of cognitive abilities that are assessed in conventional neuropsychological evaluations. While standard neuropsychological tests can provide an assessment of the ability to integrate information and draw inferences, the materials are based upon invariant, categorical dimensions. The Wisconsin Card Sort Test (Heaton, Chelune, Talley, Kay, & Curtiss, 1993), for example, assesses the ability to recognise and shift between relations based on form, colour, and number. In contrast, social information processing requires the detection of individual differences and changing contingencies. Indeed, while the goal-directed regulation of cognition has been associated with the dorsolateral convexities of the prefrontal lobes (e.g., see Stuss & Knight, 2002), many facets of social perception, including emotion perception, the ability to make socially relevant judgements (such as eye gaze direction, attractiveness, trustworthiness), and “theory of mind” (ToM) judgements, that is, inferences about another's beliefs and intentions, have been consistently attributed to overlapping systems that include the orbitomedial and ventral frontal lobes, as well as intimately related limbic structures including the anterior cingulate gyrus and the amygdala (Adolphs, 2001). Consequently, conventional neuropsychological tests, even those that are thought to be sensitive to executive impairments following frontal lobe injury, fail to detect deficits in the social realm (Saver & Damasio, 1991).
The Awareness of Social Inference Test (TASIT) (McDonald, Flanagan, & Rollins, 2011) was developed in order to assess key aspects of social perception. In this article, the test will be described, and evidence for its reliability and validity across a range of normal and clinical groups will be reviewed. Prior to this, it is important to briefly consider several core constructs in social perception that underpin TASIT, including emotion perception, ToM reasoning, and conversational inference.
Emotion perception refers to the ability to identify the emotion of others based upon facial and vocal expression. Poor emotion perception has been repeatedly linked to ventromedial and orbitofrontal regions (Angrilli, Palomba, Cantagallo, Maietti, & Stegagno, 1999; Fernandez-Duque & Black, 2005; Hornak et al., 2003; Hornak, Rolls, & Wade, 1996; Keane, Calder, Hodges, & Young, 2002; Mah, Arnold, & Grafman, 2005). Perception of negative emotions is more impaired than positive emotions in a wide variety of psychiatric and neurological disorders, including TBI (Croker & McDonald, 2005), schizophrenia (Mandal, Pandey, & Prasad, 1998), FTD (Fernandez-Duque & Black, 2005), Alzheimer's disease (AD) (Kohler et al., 2005), and stroke (Braun, Traue, Frisch, Deighton, & Kessler, 2005). Differential impairment in the processing of negatively valenced emotions following focal lesions has led to the view that it is the negative emotions, especially high arousal emotions such as anger and fear, that are mediated by ventral and medial frontal systems, including the amygdala (Adolphs, 2002).
ToM refers to the capacity to attribute mental states, such as thoughts, beliefs, desires, and intentions to others, and is pivotal to the ability to make sense of social behaviour and interpersonal communication. ToM was originally conceptualised as a core deficit in people with autism spectrum disorders. Individuals with high-functioning autism or Asperger's syndrome may have normal IQ, and have fluent and articulate speech but are, nonetheless, pedantic and over-literal (Happe et al., 1996), fail to interact normally in conversation, often talk at length on obscure or inappropriate topics (Ozonoff & Miller, 1996), have inappropriate non-verbal communication, and have poor adherence to social rules (Bowler, 1992). They also fail to appreciate how utterances are used to convey information in a socially appropriate manner (Surian, Baron-Cohen, & Van der Lely, 1996), and misinterpret both metaphor and irony (Happe, 1993).
Deficits in ToM have been reported in many clinical conditions that similarly lead to poor interpersonal skills and communicative behaviour, including schizophrenia (Brune, 2005a, 2005b), TBI (Bibby & McDonald, 2005; Dennis, Purvis, Barnes, Wilkinson, & Winner, 2001; Milders et al., 2003; Milders, Ietswaart, Crawford, & Currie, 2006, 2008; Santoro & Spiers, 1994; Stone, Baron-Cohen, & Knight, 1998; Turkstra, Dixon, & Baker, 2004), and FTD (Gregory et al., 2002; Keane et al., 2002; Rosen et al., 2004; Snowden et al., 2003).
The neurological basis for ToM deficits is unclear. On the one hand, there is some evidence that ToM is a specialised, modular facet of social cognition (Havet-Thomassin, Allain, Etcharry-Bouyx, & Le Gall, 2006; Rowe, Bullock, Polkey, & Morris, 2001) due to the absence of a clear relation between measures of executive function and ToM performance in adults with focal frontal lesions (Rowe et al., 2001). Further, ToM deficits have been reported to be most severe following ventromedial lesions (Shamay-Tsoory, Tomer, & Aharon-Peretz, 2005) and not associated with dorsolateral frontal lesions, which commonly give rise to executive impairment. However, other works suggest that problems with general inference making and working memory also affect ToM tasks (Bibby & McDonald, 2005; Martin & McDonald, 2005), and this is increasingly likely as the task becomes complex.
Conversational inference refers to the meaning that is derived from understanding language in context. In everyday communication, there are almost always other meanings being communicated than apparent from the words themselves. Indeed, often what a person means is not what they have said literally. Their true meaning may be signalled by their facial expression, tone of voice, gestures, body language, or the context in which the words are spoken. For example, the retort by a woman to her friend: “No! Of course you haven't put on weight. You're as slim as ever!” could be meant sincerely and matter-of-fact, or it could be intended as a comforting white lie to disguise the real state of affairs, or even as a sarcastic rebuke to emphasise the fact that the opposite is true. Her real meaning will be apparent from her demeanour and the context. Understanding conversational inference can be conceptualised as a higher order social perceptual skill as it relies upon the ability to integrate language with a number of processes, especially emotion perception and complex ToM judgements (McDonald & Flanagan, 2004; Shamay-Tsoory et al., 2005). The ability to interpret conversational inferences such as sarcasm has, again, been linked to frontal cerebral systems (Rankin et al., 2009; Shamay, Tomer, & Aharon-Peretz, 2002; Shamay-Tsoory et al., 2005).
TASIT was designed to provide an ecologically valid assessment of these core areas of social perception. It was designed to assess the ability to interpret emotional expression, as well as other verbal and non-verbal signals in order to make judgments about the mental and emotional state of the speakers and the meaning of their conversations. The test comprises a series of videotaped vignettes employing professional actors. It has three parts, each with alternate forms: (1) Emotion Evaluation Test; (2) Test of Social Inference (Minimal); and (3) Test of Social Inference (Enriched).
Part 1: Emotion Evaluation Test
Part 1 of TASIT (with alternate forms A and B) comprises 28 video vignettes of professional actors portraying one of the six basic emotions that are widely accepted as universal across cultures: happiness, sadness, anger, fear, revulsion (disgust), and surprise (Ekman, 1973), as well as neutral. All scripts are ambiguous monologues or dialogues devoid of specific emotional content.
Unlike conventional tests of emotion that use static photographs (e.g., the Facial Expressions of Emotion—Stimuli and Tests (Young et al., 2002) or even drawings (e.g., Palermo, Pasqualetti, Barbati, Intelligente, & Rossini, 2006; Prodan, Orbelo, Testa, & Ross, 2001), the vignettes in TASIT were designed to emulate real emotional expressions as they occur dynamically over time. The actors in TASIT were trained in a method-style of acting that requires the actor to elicit a real emotion in him or herself rather than to “indicate” or fake the required emotion as is usually the case in conventional emotion stimuli. In this way, a skilled actor is able to portray a spontaneous kaleidoscope of reactions—including “playing against” or coping (e.g., trying to reign in an emotion or display another emotion in the service of social acceptability) as often occurs in real life. Thus, Part 1 provides a series of dynamic, complex but naturalistic (i.e., ecologically valid) items that are able to capture the nature of real emotional expression more fully than conventional stimuli. Recognition is assessed by asking participants to select the correct category label from a choice of seven (the six emotions plus neutral).
Part 2: Social Inference (Minimal)
Part 2 of TASIT looks at situations in which the viewer is required to determine the speaker's meaning and intentions based upon the dialogue, emotional expression, and other paralinguistic cues. In some situations, these elements are convergent and complementary, openly communicating the speaker's intention. In other situations, the elements may be incongruous or even contradict one another. For example, a speaker may say “Great job!” while at the same time rolling his/her eyes, shaking his/her head, and adopting a closed, aggressive pose. Only by integrating all the elements within the context of what has happened enables the viewer to recognise that the speaker is being sarcastic and that his/her intended message is in fact directly opposite to the words he/she is using. In Part 2 of TASIT, there is no external or additional information provided to the viewer to assist in making this interpretation. There are three kinds of exchanges: sincere, where the speaker means what he/she says (five items); sarcastic, where the same ambiguous script is enacted in such a way that it is clear the speaker is sarcastic, meaning something quite different to that literally asserted (five items); and paradoxical sarcasm, where the script literally makes no sense (e.g., “Have you got your passport,”“No, I tore it up and threw it in the bin,”“Oh good then”) unless it is assumed one speaker is being sarcastic (five items).
Part 3: Social Inference (Enriched)
Part 3 of TASIT uses another series of vignettes, all of which entail a message that is literally untrue. Half of these, like Part 2, represent sarcastic exchanges, that is, the literal message is contrary to the actual message the speaker wishes to convey. The remainder uses the same scripts as the sarcasm vignettes, but in each one, the speaker is lying, usually to be kind, that is, he/she intends the literal message to be accepted but it is contrary to what he/she believes. Part 3 of TASIT is enriched by the inclusion of additional evidence, either a visual edit, for example, showing a full dinner plate as the parents discuss whether their son has eaten his dinner, or by a prologue between two speakers where the “lying” or “sarcastic” actor reveals his/her true thoughts or feelings to a third party—for example, immediately prior to speaking to his/her corpulent friend, the speaker is observed confiding to a third person that he/she thinks his/her friend is indeed putting on weight.
In order to discriminate the different exchange types in Parts 2 and 3, examinees need to draw conclusions about (1) the emotional state of the speakers; (2) their beliefs and knowledge about the situation (first order ToM); (3) what the speaker intends their conversational partner to think or feel as a result of the interaction (second order ToM); and (4) whether they want the literal or non-literal meaning of their message to be believed. For each vignette in Parts 2 and 3, there are four response probes that tap these different facets of comprehension. For example, after watching Ruth telling her friend that he has not put on weight, the participant is requested to answer Yes or No to the following: Does Garry seem happy with Ruth? (emotion); Does Ruth think he has put on weight? (first order ToM); Is Ruth trying to make him believe he hasn't put on weight? (second order ToM); Is she trying to say he has put on weight? (meaning).
Can a Test of Social Perception Meet Psychometric Standards of Reliability and Validity?
While high levels of reliability and validity are expected of conventional neuropsychological tests, the idea that tests of social perception can also meet such standards is relatively novel. Indeed, TASIT stands alone as a test of social perception that has been developed to the stage that it has reliability and validity statistics or normative data. One important way in which TASIT differs from many tests of neuropsychological function with normative data is that it was designed as a criterion referenced test. English speakers were expected to perform near ceiling on all subtests. This has generally been found to be true. Normative data for TASIT (McDonald, Flanagan, Rollins, & Kinch, 2003), based on a total pool of 279 adult volunteers drawn from local sports and bible clubs, the police training academy, the navy and the university, found overall scores on Part 1 (Form A) to be 24.9 out of 28 (standard deviation (SD) = 2.1); Part 2, 54.1 out of 60 (SD = 4.3); and Part 3, 55.6 out of 64 (SD = 4.8). In addition, total scores on the alternative forms of Parts 1, 2, and 3 were found to be statistically equivalent (McDonald et al., 2003).
The near ceiling performance of normal adults on TASIT limits the ability to report reliability estimates using non-clinical participants. Instead, test–retest and alternate forms reliability have been estimated based upon the more variable performance of adults with severe, chronic TBI (McDonald et al., 2006). Test–retest of Form A of TASIT (32 participants) estimated with an interval of 1 week ranged from r = 0.74–0.88. No practice effects emerged. Alternate forms reliability based on 38 participants given Form A (or B) on the first occasion and called back to do the second form between 5 and 26 weeks later ranged from 0.62 to 0.83 (McDonald et al., 2006). With the exception of alternate forms reliability for Part 2, these values all indicated reliability in the moderate to high range (0.7–0.9).
Convergent and divergent validity
Because TASIT was designed as a criterion referenced test such that normal adults with a range of levels of social skills were expected to perform well, it was anticipated that performance should be relatively unaffected by differences in demographic factors such as age and education. In general, this has been found to be true in non-clinical populations (McDonald et al., 2003), although some associations did emerge. While neither age nor education correlated with Form A: Parts 1, 2, and 3; or Form B: Parts 2 and 3, there was significant shared variance between these two factors and Form B, Part 1 (16% and 27%, respectively), indicating that the younger, better educated people did better. No association was found between performance on Part 1 (Form A or B) and intelligence, but both Parts 2 and 3 (both forms) were significantly associated with IQ (13–24%) (McDonald et al., 2003).
Construct validity is also established when it is demonstrated that a test (1) correlates with other tests that measure related constructs (convergent validity) and (2) does not correlate with tests that measure dissimilar constructs (divergent validity). The variable performance of adults with TBI on TASIT provides a good basis to examine convergent and divergent validity, and this has been reported using a pool of 116 adults with TBI (McDonald et al., 2006).
Because the material in TASIT is complex and dynamic, performance is likely to be affected by deficits in information processing and working memory (Bibby & McDonald, 2005; Channon & Crawford, 2000; Dennis & Barnes, 1990; Martin & McDonald, 2005). In addition, the ability to learn and retain information, and the ability to think flexibly (executive function) are likely to contribute to understanding TASIT. Indeed, within this clinical sample, all parts of TASIT were significantly correlated with IQ (Wechsler Test of Adult Reading, r: 0.26–0.50), information processing speed (e.g., Wechsler Adult Intelligence Scale (WAIS)-III Symbol Search, r: 0.32–0.54), and working memory (e.g., WAIS-III Digit Span, r: 0.25–0.30). Within the domains of new learning and executive function, it was anticipated that more socially relevant tasks would show association with TASIT (convergent validity), whereas non-social tasks would not (divergent validity). To a great extent, this pattern was upheld. New learning of socially relevant information, specifically stories and faces, were associated with all aspects of TASIT (e.g., Wechsler Memory Scale (WMS)-III Logical Memory 1, r: 0.33–0.39; WMS-III Faces, r: 0.42–0.69), while less socially relevant new learning such as geometric shapes and word lists were not (with the exception of an association between Verbal Paired Associates and Part 2, r: 0.31, Parts 1 and 3: r: 0.13–0.18 (ns)). Within the executive sphere, all facets of TASIT were significantly associated with WAIS-III Similarities (r: 0.29–0.49), in line with expectations as this task taps social reasoning. TASIT was also associated with the geometric problem-solving task Matrix Reasoning (r: 0.66–0.78), which was not anticipated. Other non-social problem-solving tasks, such as Wisconsin Card Sorting Test (WCST) and generating words in phonemic categories (Controlled Oral Word Association Test (COWAT)) showed no association with TASIT (WCST, r: 0.09–0.19; COWAT, r: 0.12–0.26). As an additional examination of convergent validity, TASIT correlates with social perception measures taken from the experimental literature (McDonald et al., 2006). Specifically, all parts of TASIT correlated with emotional identification using faces from the Ekman and Friesen series (Ekman & Friesen, 1976) (r: 0.37–0.69). Part 2 of TASIT also correlated with comprehension of second-order ToM stories (Bibby & McDonald, 2005) (r: 0.68) but not with simple first-order ToM stories, or comprehension of stories that contain physical inferences only.
Performance of specific clinical groups. An additional method for establishing construct validity is to examine test performance in a group known to have deficits in the relevant constructs. In the case of TASIT, it is expected that clinical populations who do not suffer from language impairment but do experience deficits in social perception will have impairment in the recognition of emotions (Part 1) and interpretation of conversational remarks meant to be taken non-literally, that is, sarcasm (Parts 2 and 3). In contrast, they are not expected to have difficulty with sincere remarks (Part 2). Problems with making ToM judgements, especially higher order ToM, such as judging intention, are also anticipated in people with disorders of social perception. There have now been a number of reports published on different clinical populations that provide additional evidence as to the validity of TASIT to measure clinical social perception disorders.
TBI. TBI, typically sustained as a result of motor vehicle accidents, falls, and assaults, causes diffuse damage throughout the cerebrum and multifocal lesions concentrated in the ventrolateral frontal lobes, inferior orbital cortices, and temporal lobes (Gentry, Godersky, & Thompson, 1988; Hadley et al., 1988). Adults with TBI are often egocentric and self-focused, lacking interest in other people, and displaying inappropriate humour and a blunt manner (Crosson, 1987; Flanagan, McDonald, & Togher, 1995; Levin, Grossman, Rose, & Teasdale, 1979; McDonald, Flanagan, Martin, & Saunders, 2004). They have also been reported to find it difficult to identify the source of interpersonal conflict (Kendall, Shum, Halson, Bunning, & Teh, 1997) or otherwise interpret non-verbal interpersonal interactions (Bara, Cutica, & Tirassa, 2001; Cicerone & Tanenbaum, 1997). People with TBI are poor at recognising emotions in others (Green et al., 2004; Hopkins, Dywan, & Segalowitz, 2002; Ietswaart et al., 2008; Jackson & Moffatt, 1987; Milders et al., 2003), fail conventional ToM tests (Bibby & McDonald, 2005), and are poor at detecting sarcasm (Channon, Pellijeff, & Rule, 2005; Channon & Watts, 2003).
TASIT has been used as a measure of deficits in social perception following TBI in a number of studies (McDonald & Flanagan, 2004; McDonald et al., 2003, 2004; McDonald & Saunders, 2005) that attest to its construct validity. In general, TBI participants have been found to be poorer than their non-brain-damaged counterparts on all Parts of TASIT. In each group studied, the people with TBI were more variable than the matched non-clinical control group. This is expected given that TBI is a heterogeneous condition, and not all speakers with TBI are expected to suffer deficits in social perception. It also indicates that TASIT is potentially sensitive to a range of competencies in social perception in this clinical group.
Within task, variability was also found. In one study (McDonald & Flanagan, 2004), the TBI group had difficulty across all emotion categories, while in another (McDonald et al., 2003), specific problems with anxiety, disgust, and neutral expressions were reported. Difficulty with the neutral expressions may reflect the generally ambiguous nature of these vignettes. Failure to recognise anxiety and disgust is, however, consistent with independent reports that people with TBI have particular difficulty recognising negative emotional expressions (Croker & McDonald, 2005; Jackson & Moffatt, 1987; Prigatano & Pribam, 1982). Within Parts 2 and 3, the TBI participants were poor on all types of exchanges except sincere exchanges (Part 2) and lies (i.e., insincere remarks that are intended to be taken literally; Part 3). This pattern is consistent with other empirical accounts of communication disorders in TBI that report that they have little difficulty with literal meanings compared with non-literal or inferred meanings that rely upon context for proper interpretation (Dennis, Purvis, Barnes, Wilkinson, & Winner, 2001; McDonald, 1992; Pearce, McDonald, & Coltheart, 1998). Lies may pose little difficulty as the liar intends his/her message to be taken literally. It was found that emotion perception (Part 1) per se was not related to ability to understand inference (McDonald & Flanagan, 2004) in this group, a finding that mirrors earlier work using different methods (McDonald & Pearce, 1996). In contrast, ability to accurately understand speaker intention (higher order ToM) was related (McDonald & Flanagan, 2004). Collectively, these findings provide evidence of the construct validity of TASIT as a measure of social perception.
Schizophrenia. Social cognition deficits have been commonly reported in people with schizophrenia, including impaired recognition of facial and prosodic expressions of emotion, especially negative emotions (Edwards et al., 2002) and deficits in ToM judgements (Brune, 2005a, 2005b). Deficits in social perception as measured by TASIT within this population are similar to those reported following TBI. Adults with schizophrenia have been reported to have difficulty recognising negative emotions from Part 1 of TASIT (Sparks, McDonald, Lino, O'Donnell, & Green, 2010) and to have difficulty with understanding sarcastic exchanges, while comprehension of sincere exchanges is normal (Kern et al., 2009; Kosmidis, Aretouli, Bozikas, Giannakou, & Ioannidis, 2008; Sparks et al., 2010). Problems with understanding lies have also been reported (Kosmidis et al., 2008; Sparks et al., 2010), although not always (Kern et al., 2009). Poor TASIT scores are significantly associated with positive symptoms (Kern et al., 2009), reporting of empathic distress and reduced recreational functioning (Sparks et al., 2010).
Although TASIT is sensitive to social perception deficits in schizophrenia, it is not sensitive to schizotypy in undergraduate students (Jahshan & Sergi, 2007). As TASIT was designed specifically to be a criterion referenced test, achieving high scores in healthy adults with a range of social skills, it was not designed to measure individual differences in normal speakers. The fact that it does not differentiate normal speakers on the basis of personality characteristics, but does discriminate those with diagnosed clinical conditions, further attests to its clinical validity.
FTD. FTD is a degenerative condition that reflects early deterioration of frontotemporal and amygdala structures (Broe et al., 2003). The behavioural variant (bvFTD), in particular, is heralded by changes to social behaviour and social cognition. It is well established that people with FTD have difficulties in emotion perception and ToM judgements (Gregory et al., 2002; Keane et al., 2002; Rosen et al., 2004; Snowden et al., 2003). Studies using TASIT have confirmed social perception deficits. Specifically, people with bvFTD demonstrate poor ability to detect emotions on Part 1 of TASIT (Kipps, Nestor, Acosta-Cabronero, Arnold, & Hodges, 2009; Rankin et al., 2009), especially negative emotions (Kipps et al., 2009). They also demonstrate impaired ability to comprehend sarcasm on Part 2 (Kipps et al., 2009; Kosmidis et al., 2008; Rankin et al., 2009) but are normal when processing sincere vignettes. Capacity to understand sarcasm (Part 2) is related to capacity to identify emotions (Part 1) in this population (Kipps et al., 2009; Rankin et al., 2009), especially negative emotions (Kipps et al., 2009).
Importantly, performance on TASIT has been shown to differentiate patients with bvFTD who have significant cerebral pathology from those who do not, despite essentially similar clinical presentations in the two subgroups (Kipps et al., 2009). This differentiation was not achieved using a test of more global cognitive function. The utility of TASIT to discriminate between these variants is especially valuable, as there are a number of patients with clinically diagnosed bvFTD who do not progress, leading to doubts regarding the underlying neuropathology. To date, TASIT is the only objective test that has proven it is possible to differentiate progressive from non-progressive bvFTD. Poor performance on TASIT was associated specifically with damage to the lateral orbitofrontal cortex, insular and the temporal poles, especially the right (Kipps et al., 2009).
AD. AD is a degenerative condition that has quite a different profile to FTD. In AD, failing memory and learning are considered the early hallmarks, although broad-ranging cognitive deficits affecting executive processes, language, and perception may develop simultaneously (Backman, Jones, Berger, Laukka, & Small, 2004). Degenerative changes occur throughout the neocortex and hippocampus, as well as the amygdala, and basal nucleus of Meynert (Wenk, 2003). AD differs from FTD in that changes to social perception are not marked early in the disorder. While patients are variable on conventional tests of labelling (static) emotions (Burnham & Hogervorst, 2004; Hargrave, Maddock, & Stone, 2002; Lavenu, Pasquier, Lebert, Petit, & Van der Linden, 1999), normal performance has been reported (Burnham & Hogervorst, 2004; Lavenu et al., 1999), especially when AD patients have been screened for language deficits (Cadieux & Greve, 1997). Where studies have been conducted on TASIT performance with this population (mainly used as a comparison group to FTD), performance is also relatively preserved. TASIT emotion perception (Part 1) has been shown to be only mildly affected (Henry et al., 2008; Kipps et al., 2009) or normal (Rankin et al., 2009), while the ability to understand sarcasm (Part 2) has been found to be intact (Kipps et al., 2009; Rankin et al., 2009; Rapp & Wild, 2011).
Right hemisphere lesions. Patients with lesions limited to the right hemisphere (as a result of cerebrovascular accident, etc.) typically do not suffer aphasia. Nevertheless, it has long been established that they do demonstrate disorders in their ability to understand non-literal language such as metaphor (Winner & Gardner, 1977), proverbs (Hier & Kaplan, 1980), idioms (Van Lancker & Kempler, 1987), and jokes (Birhle, Brownell, Powelson, & Gardener, 1986). They are also less able to use the information provided about the mood of an individual (angry vs happy) to interpret whether a conversational remark has a sarcastic or joking intent (Brownell, Carroll, Rehak, & Wingfield, 1992; Kaplan, Brownell, Jacobs, & Gardner, 1990). It is also the case that patients with focal lesions to the right hemisphere demonstrate significant impairment in the recognition of emotion from both faces and voice relative to those with left hemisphere lesions and healthy controls (Kucharska-Pietura, Phillips, Gernand, & David, 2003). Once again, work using TASIT has demonstrated its sensitivity to social perception deficits in this population with such patients demonstrating poor understanding of emotions (Part 1), especially negative emotions (Fournier, Calverley, Wagner, Poock, & Crossley, 2008) and poor capacity to comprehend sarcasm (Blake, 2009; Fournier et al., 2008), with better performance on lies (Fournier et al., 2008). Left hemisphere pathology, while compromising language function, lead to relatively normal performance on TASIT (Fournier et al., 2008).
Validity can also be demonstrated via the ability of the test to predict real-world function (ecological validity). To this end, we examined the association between performance on TASIT and social behaviour in 21 adults with severe, chronic TBI (Post Traumatic Amnesia (PTA) ranging from 3 to 240 days) (McDonald et al., 2004). Following their assessment on TASIT, the participants with TBI were video-recorded during a spontaneous encounter with a confederate. The video-recordings were rated by two raters using scales from the Behaviourally Referenced Rating System of Intermediate Social Skills—Revised (Wallander, Conger, & Conger, 1985). These were the Personal Conversational Style (inter-rater agreement r = 0.71), with three subscales rating self-disclosure, use of humour, and social manners (e.g., politeness, use of compliments, interruptions), and the Partner-Directed Behaviour Scale (inter-rater agreement r = 0.77), with three subscales rating the use of verbal reinforcements, egocentric behaviour, and partner involvement behaviour (e.g., getting the other person to talk about himself/herself). Ecological validity was demonstrated as the Use of Humour subscale was significantly correlated with Part 1 (r = 0.46), Part 2 (r = 0.58), and Part 3 (r = 0.67). Social manners, Use of Reinforcers, Egocentric Behaviour, and Partner Involvement were each significantly correlated with Part 3 (r = 0.57, 0.57, 0.77, 0.49 respectively). In this study, we also asked relatives of the participants with TBI to complete a questionnaire regarding everyday social behaviour: the Social Performance Survey Schedule (Lowe & Cautela, 1978). No association was found between TASIT and either positive or negative behaviours as assessed by the SPSS. In retrospect, this nil finding was not surprising as the questionnaires tapped a broad range of behaviours. This, combined with the range of chronicity of the people with TBI, the variable nature of their injuries, and the variable nature of the relationship of the informant to the person with TBI, introduced a great deal of variance over and above that measured by TASIT.
In two independent studies of people with severe TBI, performance on items from Part 1 of TASIT was associated with close other perception of communication competence as measured by the Latrobe Communication Questionnaire (Watts & Douglas, 2006), and social integration as measured by the Revised Craig Handicap Assessment and Reporting Technique (Knox & Douglas, 2009; Mellick, Walker, Brooks, & Whiteneck, 1999). Similarly, poor performance on TASIT has been related to reduced real-world functioning in social and work activities in people with schizophrenia (Mancuso, Horan, Kern, & Green, 2011).
In conclusion, the domain of neuropsychological assessment needs to expand to encompass social cognition. Recent research suggests that disorders of social cognition are a major feature of many clinical conditions, and yet these are not easily detected using standard neuropsychological measures. While the ability to understand complex social information will partially rely on generic skills, including information processing speed, working memory, and executive abilities, there are other aspects to social cognition that are unique and, arguably, dissociable. TASIT has evidence for reliability and validity, demonstrating that measures of social perception can stand up to psychometric standards. Of course, TASIT has a long way to go to achieve the kinds of high levels of reliability seen in well-established tests, such as the Wechsler Scales. Nonetheless, the use of multiple choice labels for the emotion recognition tasks and Yes/No answers for Parts 2 and 3 ensures that response can be unambiguously coded, which increases reliability of rater scoring. By carefully constructing items, it is also possible to provide stimuli that are interpretable by the majority of normal adults, yielding a test that shows up clinical deficits in social perception. While TASIT was designed as a criterion referenced test and should, therefore, be relatively unaffected by individual differences, it was the case that some subtests were sensitive to age, education, and intelligence. Consequently, performance on TASIT, especially Parts 2 and 3 which require good language processing, needs to be interpreted in this light. Further studies that can examine the extent to which TASIT predicts specific social dysfunction independently of these variables would be useful.
As demonstrated in the arena of FTD, measures of social perception such as TASIT have important diagnostic utility, proving to be more sensitive to clinical variants of certain kinds of neuropathology than standard neuropsychological batteries. TASIT also plays an important role in revealing targets for remediation. The use of naturalistic, complex video vignettes provides a direct analogue to everyday situations in which people with clinical disorders have to function. This, combined with systematic theory-driven categories of stimuli and response probes, enables clinicians to have a clear grasp of the kind of neuropsychological mechanism affected. Accurate assessment translates to clear specification of treatment goals.
Furthermore, treatment in the area of social perception is gaining momentum with remediation research, like assessment, turning squarely into the area of social cognition. Results are promising, especially for emotion perception, even in those with long-standing, developmental impairments, such as autism (Bolte et al., 2006; Solomon, Goodlin-Jones, & Anders, 2004) and intellectual disability (McAlpine, Singh, Kendall, & Hampton, 1992; McKenzie, Matheson, McKaskie, Hamilton, & Murray, 2000). Emotion remediation and more general social cognition remediation are also effective for people with schizophrenia (Combs et al., 2007; Combs, Tosheva, Wanner, & Basso, 2006; Frommann, Streit, & Wolwer, 2003; van der Gaag, Kern, van den Bosch, & Liberman, 2002; Penn & Combs, 2000; Wolwer et al., 2005) and TBI (Bornhofen & McDonald, 2008a, 2008b).
Because TASIT has alternate forms, it can play a valuable role, not only in assessment but also treatment itself. For example, Form A can be used for both pre- and post-treatment evaluation (Bornhofen & McDonald, 2008a, 2008b), while Form B, with similar but not identical vignettes, can be used as practice stimuli during therapy. Because TASIT has several parts focusing on different aspects of social perception (emotion, ToM, conversational inference), and is audiovisual providing the option to present it with sound alone (by obscuring the picture), visual only (by turning the sound off) or audiovisual, it provides a range of treatment options.
It is hoped that the success in the development of TASIT will be a clarion call for greater research into other assessment instruments that can detect and characterise deficits in social information processing. One important new direction is in the expansion of the kinds and types of emotion that are assessed. In TASIT, six emotions (happy, surprised, angry, fear, sad, and disgust) are represented, reflecting the basic emotions thought to be universal across cultures (Ekman, 1973). However, the range of these emotions is very limited, not only in terms of the number of positive emotions provided but also in the provision of more “social emotions,” such as pride and embarrassment. There is also a need to examine cross-cultural differences in sensitivity to a range of emotional expressions. In addition, as the literature expands to conceptualise more facets of social perception (e.g., judgements about physical attractiveness (Kampe, Frith, Dolan, & Frith, 2001; O'Doherty et al., 2003); sexual orientation (Ishai, 2007); trustworthiness (Adolphs, Tranel, & Damasio, 1998; Winston, Strange, O'Doherty, & Dolan, 2002); intimacy (Mah, Arnold, & Grafman, 2004); and personality (Heberlein, Adolphs, Tranel, & Damasio, 2004)), so too, demands grow for sensitive, reliable, clinical measures in this vital domain of human performance. While work with TASIT has demonstrated that social perception can be assessed in a reliable manner, it is also important to be cognizant of the inherent constraint on the development of social perception tests. There is already a very healthy debate regarding the extent to which standard cognitive tests are “culture free” with clear evidence that they are not. Social perception, by its very nature, is culturally situated. This makes the development of social perception tests particularly challenging. It needs to be clearly understood that the validity of the tasks, even more so than conventional cognitive measures, is maximal for the cultural cohort of the test-makers. This does not negate the value of such measures but it does emphasise the need for research into different cultural groups in order to both validate existing measures such as TASIT and to develop others that are suitable for diverse cultural settings.