| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
RESEARCH ARTICLE |
School of Psychology, University of New South Wales, Sydney, Australia.
Address correspondence to Phoebe Bailey, School of Psychology, University of New South Wales, Sydney NSW 2052, Australia. E-mail: pbailey{at}psy.unsw.edu.au
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Disinhibition • Age differences • Theory of mind
Cognitive empathy, perspective-taking, and theory of mind (ToM) are overlapping constructs that refer to understanding another's beliefs and desires, and thus they are essential prerequisites for successful social interaction (Eslinger, 1998
). It has been argued that these social reasoning skills require a regulatory mechanism that allows one to adopt the subjective perspective of another (Decety & Jackson, 2004; Eslinger, 1998; German & Hehman, 2006). This is because executive selection processes are necessary in order to decide on the most appropriate mental state attribution from a number of potential candidates (German & Hehman). In addition, the self-perspective is regarded as the cognitive "default mode," driven by the automatic link between perception and action (Decety et al., 1997). Therefore, to evaluate another's perspective, some form of active inhibitory mechanism must regulate the prepotent self-perspective.
Suppressing alternative perspectives in order to take another's perspective is consistent with the conception of inhibition by Miyake, Friedman, Emerson, Witzki, and Howerter (2000) in which inhibition is an intended or deliberate process. That is, it is the process of stopping a response that is relatively prepotent, dominant, or automatic. This is in contrast to the concepts of reactive inhibition and inhibition in typical spreading activation models (Logan, 1994), which lack the deliberate intention and control that is central to inhibition in belief–desire reasoning.
Consistent with theoretical models (Decety & Jackson, 2004; German & Hehman, 2006), limited inhibitory control in childhood has been linked to a reduced capacity for ToM (e.g., Carlson, Moses, & Claxton, 2004). Normal aging is also associated with reduced inhibitory control (Braver et al., 2001; Kane, Hasher, Stoltzfus, Zacks, & Connelly, 1994), and therefore inhibitory losses may be one mechanism underpinning a tendency to grow less empathic with age. Indeed, whereas Happé, Winner, and Brownell (1998) found improved ToM in late adulthood, and other researchers found no significant age-related differences in this capacity (Keightley, Winocur, Burianova, Hongwanishkul, & Grady, 2006; MacPherson, Phillips, & Della Sala, 2002), most studies have identified age-related deficits in ToM (German & Hehman; Maylor, Moulson, Muncer, & Taylor, 2002; McKinnon & Moscovitch, 2007; Phillips, MacLean, & Allen, 2002; Sullivan & Ruffman, 2004), perspective-taking (Ligneau-Hervé & Mullet, 2005), and cognitive empathy (Bailey, Henry, & von Hippel, in press). This suggests that, as we age, it becomes more difficult to see things from someone else's point of view.
To our knowledge, only four studies to date have investigated whether the capacity for executive functioning predicts age-related changes in belief–desire reasoning: the one by Saltzman, Strauss, Hunter, and Archibald, (2000); the one by Maylor and colleagues (2002, Experiment 2); the study by German and Hehman (2006); and that by McKinnon and Moscovitch (2007). Although Saltzman and colleagues found some evidence of a link between impaired ToM and executive functioning, as measured by the California Card Sorting Task, verbal fluency, and 5-point fluency, only eight older participants contributed to these analyses. In a larger study, Maylor and associates found that although ToM was significantly reduced in older adults relative to younger adults, this deficit was independent of reactive and spontaneous flexibility, as indexed by the Wisconsin Card Sorting Test and excluded letter fluency, respectively. Thus, executive functioning as operationalized by these measures was not related to deficient ToM and could not explain the age effects observed. Nevertheless, Maylor and associates noted that other aspects of executive functioning may be more critical to empathic processing.
Indeed, it has recently been argued that older adults are disproportionately impaired on ToM measures that impose greater demands on executive resources (German & Hehman, 2006; McKinnon & Moscovitch, 2007). Consistent with this possibility, German and Hehman found that increased executive demands in four different belief–desire reasoning tasks were associated with decreased accuracy, particularly among the older age group. In addition, these researchers found cognitive inhibitory control to contribute significantly to the variance in response times on these tasks. Similarly, McKinnon and Moscovitch compared the performance of younger and older adults on first- and second-order false-belief tasks. Whereas the former requires the adoption of the perspective of an individual character, the latter requires the adoption of two perspectives simultaneously and thus is argued to place greater demands on executive functioning. Relative to younger adults, older adults were impaired only on the inference of second-order false beliefs, and McKinnon and Moscovitch construed this as evidence that increasing executive demands may result in decreased ToM in late adulthood.
However, despite theoretical and empirical evidence that age-related declines in belief–desire reasoning are associated with cognitive disinhibition (Decety & Jackson, 2004; German & Hehman, 2006), to our knowledge no study to date has directly tested whether disinhibition mediates these age-related declines. In addition, no study involving older adults has previously assessed whether cognitive inhibition (i.e., nonsocial inhibition) and inhibition of the self-perspective (i.e., social inhibition) converge within the context of belief–desire reasoning. To address these issues, in addition to testing how inhibitory control related to performance on a standard test of ToM (i.e., the Revised Eyes Test; see Baron-Cohen, Wheelwright, Hill, Raste, & Plumb, 2001), in the current study we also used a manipulation previously applied to a case study of an individual, known as patient WBA, with right prefrontal and temporal brain injury (Samson, Apperly, Kathirgamanathan, & Humphreys, 2005). This involved the reality-known (Samson et al.) and reality-unknown (Apperly, Samson, Chiavarino, & Humphreys, 2004; Samson, Apperly, Chiavarino, & Humphreys, 2004) false belief (FB) tasks, which demand high and low levels of self-perspective inhibition, respectively. Interpretation of another's FB is a common measure of ToM, as it ensures that the correct answer cannot be determined from a participants' own perspective. Thus, it was shown that although patient WBA was capable of inferring that another person held a FB when the self-perspective inhibitory demands were low, when these demands were increased WBA could no longer infer that same FB. These findings suggest, once again, that ToM requires at least some degree of controlled inhibition of the self-perspective.
We believe the present study was the first to use this methodology to test the potential role of disinhibition of the self-perspective as a mechanism underlying reduced ToM in late adulthood. We predicted that, although older adults would perform as well as younger adults on the FB task that imposed few demands on self-perspective inhibitory control, they would be significantly impaired on the task that substantially loaded this capacity. We also predicted, in line with Phillips and colleagues (2002), that older adults would not perform as well as younger adults on the Revised Eyes Test. This measure involves the attribution of complex mental states to pictures of eyes; according to German and Hehman (2006), viewing each pair of eyes may generate a number of potential mental state attributions from which inappropriate attributions must be inhibited and the most likely selected. Thus, in addition to mediating the predicted association between age and high-inhibition FB, measures of cognitive inhibition were also expected to mediate the predicted association between age and the eye test.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). There were no significant differences in years of education for younger and older adults: younger adults, M = 14.08, SD = 1.66; older adults, M = 13.09, SD = 2.84; t(67) = 1.79, p =.08. There were no such differences for self-reported health as well (out of seven, with higher scores indicating better health): younger adults, M = 5.44, SD = 1.18; older adults, M = 5.25, SD = 0.92; t(66) = 0.75, p =.46. As is typical in aging studies, although the older adults scored significantly higher than the younger adults on a measure of crystallized intelligence (the 30-item Mill Hill Test of vocabulary; see Raven, Court, & Raven, 1988), the younger adults scored significantly higher than the older adults on a measure of fluid intelligence (the 40-item Progressive Matrices Test; see Raven, 1938). See Table 1 for descriptive and inferential statistics.
|
; Samson et al., 2004) task consists of 48 short nonverbal video clips, including 12 FB trials, 12 memory-control trials, and 24 filler trials. For each trial the participants were told that a woman would help them locate an object in one of two canisters. Figure 1(b) depicts the event sequence for the FB trials in which the woman left the room after being shown by a man where the object was hidden. Then the man switched the canisters in her absence, and she then reentered the room and pointed to the canister she believed the object was in. Participants were then asked to indicate which canister the object was in. Thus, participants were required to infer that the woman held a belief that was false (i.e., she believed the object was in the canister in which she saw it before she left the room).
|
for further details).
The reality-known (high-inhibition; see Samson et al., 2005) task consists of 48 nonverbal video clips, including 12 FB trials, 12 memory-control trials, and 24 filler trials. These were similar to the trials in the low-inhibition task except that participants were asked to indicate which of two canisters the woman would open first to find the object. Figure 1(a) depicts the event sequence for the high-inhibition FB trials. The critical difference between these and the low-inhibition FB trials was that while the woman was out of the room the participants were shown the object as it was being switched from one canister to the other. Thus, when the woman reentered the room and the participants were asked which canister the woman would open first to find the object, not only did they have to infer that the woman held a FB (i.e., she believes the object is in the canister she saw it in before she left the room), but higher demands were placed on the need to inhibit their own knowledge of the object's location. It is also important to note that whereas participants are asked directly about the woman's belief in this high-inhibition task, their understanding of the woman's belief is only indirectly inferred in the low-inhibition task (i.e., based on which canister the participant selects). The memory-control trials were exactly the same as the FB trials except that the participants were not required to infer a FB. That is, after the man removed the object, he placed it back into the same canister (rather than the alternative canister). Again, filler trials were included so that FB trials would remain a minority (see Samson et al., 2004 for further details).
We administered all trials in separate counterbalanced blocks comprising 12 trials each, including 4 FB trials, 4 memory-control trials, and 8 filler trials. Participants received feedback after each trial, thus providing an opportunity to learn from any mistakes.
We also used the Revised Eyes Test (Baron-Cohen et al., 2001) to index ToM. This measure requires participants to select which of four words best describes the thoughts or feelings expressed in 36 pictures of eyes. To ensure that a belief or intention was attributed to the person in the picture, we had our choices consist of words signifying only complex mental states, such as interested, jealous, contemplative, panicked, and embarrassed, and not basic emotions, such as happy, sad, and angry, which can be recognized without the need to attribute a belief. Higher scores indicate greater ToM.
We administered a paper version of the Stroop task (Trenerry, Crossen, De Boe, & Leber, 1989) as a measure of participants' abilities to cognitively inhibit prepotent responses. First, participants were required to read aloud 112 color words printed in congruent colors. Second, they were required to say aloud the color of 112 color words that were printed in incongruent colors (e.g., the word blue printed in green ink). An interviewer asked the participants to read these lists as quickly as possible, and a 120-second time limit was adhered to for each list. In accordance with the scoring procedure by Trenerry and colleagues, we calculated total scores as the number of items completed on the second list minus the number of errors made on that list, with higher scores indicative of superior inhibitory control. The test–retest reliability for this measure has previously been estimated to be very high (0.90; Trenerry et al.).
We also used the Hayling Sentence Completion Test (Burgess & Shallice, 1997) to index cognitive inhibition. We administered the measure in two parts. In the first part, participants were read 15 short sentences, each with the last word missing, and were asked to say a word as quickly as possible that would complete each sentence. In the second part, participants were read a similar list of 15 short sentences and were asked to say a word as quickly as possible that would not complete the sentence. Thus, whereas the first part taps the capacity for automatic response initiation, successful completion of the second part requires controlled inhibition of a prepotent "correct" response. In line with Burgess and Shallice, we constructed overall scaled scores from the addition of scaled scores derived from response times to Parts 1 and 2 separately, and errors made in Part 2, with higher scores indicative of greater cognitive inhibitory capacity. Norms for these scores are based on the initial standardization of this test with one control group and three neurological lesion groups (Burgess & Shallice). The test–retest reliability of the overall scaled score has been estimated to be 0.76 (Burgess & Shallice).
Inhibitory control functions should not be measured in isolation of other cognitive functions such as memory and processing speed (Crawford & Henry, 2005; Friedman & Miyake, 2004). Therefore, to test the possibility that any observed mediating effects of the inhibition measures may be attributable to more general features of task performance, we also assessed the following aspects of cognition: short-term memory (Forward Digit Span from the Revised Wechsler Adult Intelligence Scale, or WAIS-R; see Wechsler, 1981); working memory (Backward Digit Span from the WAIS-R); mental flexibility (alternating fluency; see Henry & Phillips, 2006); and cognitive speed (oral version of the Symbol Digit Modalities Test; see Smith, 1982).
Participants completed two sessions of testing in a counterbalanced order on separate days. In the first session, groups of approximately 10 participants were administered two blocks of the high-inhibition task, one block of the low-inhibition task, and the Matrices, Mill Hill, and Eyes tests. In the second session, individuals completed one block of the high-inhibition task, two blocks of the low-inhibition task, the Symbol Digit Test, the Digit Span tasks, the alternating fluency task, the Hayling Test, and the Stroop task.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
defines effect sizes of 0.2 as small, 0.5 as medium, and 0.8 as large. It can be seen that there were no significant age differences on the Backward Digit Span. However, the younger adults performed significantly better than the older adults on the Forward Digit Span, Symbol Digit Test, Hayling Test, Stroop Task, and alternating fluency task. Younger adults also demonstrated significantly higher levels of accuracy on the Eyes Test.
We analyzed the video task trials (FB and memory control) by using separate factorial mixed-design analyses of variance with age as the between-subjects factor (younger, older) and task type as the within-subjects factor (high-inhibition, low-inhibition). The mean and error scores for each of these trials are displayed in Figure 2. For the FB trials there was a significant main effect of age group, F(1, 67) = 7.64, p <.01, 
p2 =.10, and of task type, F(1, 67) = 5.15, p <.05, p2 =.07. Of particular interest, the Age Group x Task Type interaction was also significant, F(1, 67) = 7.69, p <.01, p2 =.10. For the memory-control trials there was a significant main effect of age group, F(1, 67) = 8.55, p <.01, p2 =.12, and of task type, F(1, 67) = 9.84, p <.01, p2 =.13, but no significant interaction, F(1, 67) =.99, p =.32, p2 =.02.
|
For the memory-control trials, two significant main effects indicate that, overall, the memory-control trials in the high-inhibition task were more difficult than those in the low-inhibition task, and older adults had greater difficulty than younger adults on both types of task. The absence of an interaction indicates that the two groups were equated in terms of relative difficulty experienced across the low- and high-inhibition memory-control trials. However, it is important to note that performance on the low-inhibition memory trials was at ceiling in the younger group, and it was approaching ceiling in the older group. Therefore, this null interaction has to be interpreted with caution.
It might be argued that the older adults' errors on the high-inhibition FB trials were indicative of advanced (rather than reduced) ToM. This is because the same actors were used to demonstrate each task. Therefore, the older adults might have believed that the female character would learn that she was being deceived and would thus select the canister that was opposite to the canister in which she last saw the object (which would be considered an incorrect response). If this were the case, the older adults would be expected to perform well at first, upon their initial encounters with the female character, and then proceed to make errors upon increased familiarity with the female character. To control for this possibility, we conducted analyses on the basis of each participant's first two responses to each type of trial in both the high- and low-inhibition tasks. However, these results did not differ from the results based on completion of all trials.
The measures of cognitive inhibition (i.e., the Hayling Test and the Stroop task) were correlated in the present study (r =.34, p <.01), but this appeared to be attributable to their common association with age. That is, the correlation was not significant among the separate age groups; it was also not significant when we conducted a partial correlation controlling for age (all ps >.73). Indeed, different measures of inhibition generally do not correlate well (Kramer, Humphrey, Larish, Logan, & Strager, 1994), and theories positing inhibition to be a unified construct have been described as overly ambitious (Friedman & Miyake, 2004). Thus, we did not construct a single measure of inhibitory control.
We conducted a series of regression analyses to test whether the associations between age (i.e., the predictor) and the measures of ToM (i.e., the criteria) were mediated by inhibitory capacity. It is important that "a given variable may be said to function as a mediator to the extent that it accounts for the relation between the predictor and the criterion" (Baron & Kenny, 1986, p. 1176). Thus, in accordance with Judd and Kenny (1981), we first regressed each putative mediator (i.e., the Stroop task and the Hayling Test) on the predictor. Second, we regressed the criteria separately on the predictor; third, we regressed the criteria separately on both the predictor and each mediator. To establish mediation, the predictor must affect the mediator in the first equation and the criterion in the second equation, and the mediator must affect the criterion in the third equation (i.e., when the predictor is also included in the model). If these criteria are met, then the effect of the predictor on the criterion will be reduced.
As shown in Figure 3(a), including the Stroop task performance in the third regression equation eliminated the effect of age on high-inhibition FB by reducing it from β = –.42, p <.001 to β = –.18, p =.21. This was a significant reduction according to a Sobel test (Preacher & Hayes, 2004; z = 2.16, p <.05), resulting in mediation. Thus, after controlling for Stroop task performance, we found that age no longer exerted a direct effect on determining a FB under conditions of high self-perspective inhibitory demands. Rather, age was associated with increased cognitive disinhibition, as indexed by the Stroop task, which in turn was associated with reduced ToM, as indexed by high-inhibition FB performance. As seen in Table 4, the Hayling Test was not a mediator of the association between age and high-inhibition FB.
|
|
To test whether mediation was selective to reduced inhibitory capacity, we entered other aspects of cognitive functioning that were also subject to age-related decline as potential mediators (i.e., the Matrices Test, Forward Digit Span, the alternating fluency task, and the Symbol Digit Test). Again, in accordance with Judd and Kenny (1981), we first regressed these potential mediators separately on the predictor (i.e., age). Second, we regressed the criteria (i.e., high-inhibition FB and the Eyes Test) separately on age; finally, we regressed the criteria separately on both the predictor and each mediator. Table 2 presents Pearson correlation coefficients that indicate the direct effects of age on these potential mediators (i.e., the first step in mediation), and on the criteria when the mediators are not included in the model (i.e., the second step in mediation). Tables 3 and 4 present the third step in establishing mediation; in other words, they show the standardized regression coefficients indicating the effects of age on the criteria and the effects of the potential mediators on the criteria, when all three variables are included in the equation. As we noted previously, to establish mediation, the mediator must continue to affect the criterion in the third equation. Thus, we can see that none of the additional measures of cognitive functioning mediated the age–ToM associations.
|
|
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; German & Hehman, 2006; Ligneau-Hervé & Mullet, 2005; Maylor et al., 2002; McKinnon & Moscovitch, 2007; Phillips et al., 2002; Sullivan & Ruffman, 2004), with older adults significantly impaired on both the Eyes Test and the high-inhibition FB trials. As we predicted, age-related difficulties were selective to the FB trials that imposed substantial demands on inhibitory control of the self-perspective. Thus, the present results provide the first evidence that we know of that disinhibition of the self-perspective contributes to an age-related decline in ToM. In addition, the demonstrated age–ToM associations were mediated by measures of cognitive inhibition, and not other aspects of cognitive functioning, including fluid intelligence, memory, mental flexibility, and cognitive speed. Furthermore, although the case study by Samson and colleagues (2005) involving lesion patient WBA demonstrated convergence of these social and nonsocial forms of inhibitory control in ToM, this may be the first study to do so in the context of normal adult aging.
It has been argued that poor performance on the high-inhibition FB trials together with good performance on the low-inhibition FB trials may indicate a serial process whereby one must inhibit one's own perspective before taking on another's perspective (Samson et al., 2005). Thus, the current results suggest that it may be the former process that is breaking down in late adulthood. Moreover, neuroimaging and lesion studies have shown that whereas the right inferior prefrontal lobe, including the inferior gyrus, is involved in the inhibition of one's own perspective (e.g., Ruby & Decety, 2004; Samson et al., 2005), the left temporal lobes are involved in the inference of someone else's perspective (e.g., Apperly et al., 2004; Ruby & Decety; Samson et al., 2004). Thus, although compared with other regions of the brain both the frontal and temporal areas are subject to disproportionate decline with age (see Phillips & Henry, 2005), the current results are consistent with the suggestion that the prefrontal cortex may be more vulnerable to age-related decline than are the temporal lobes (Raz et al., 2004).
In line with German and Hehman (2006), evidence for the importance of cognitive inhibitory control in empathic inference comes from the mediational analyses of the present study. It is of note that older age was not directly associated with reduced high-inhibition FB performance. Rather, older age was associated with reduced cognitive inhibitory capacity, which in turn was related to decreased accuracy on FB tasks that placed high demands on inhibitory control of the self-perspective. Additionally, the direct effect of aging on the Eyes Test performance was attributed in part to reduced cognitive inhibition. Further, although German and Hehman found that reductions in processing speed were also associated with age-related differences in belief–desire reasoning, other aspects of cognitive functioning did not mediate the age–ToM associations in the current study. Nevertheless, it is important to note that successful completion of the Hayling Test of controlled inhibition, which mediated the age–Eyes Test association, also relied in part on capacity for automatic response initiation.
It is of note that whereas the Stroop task mediated the association between age and high-inhibition FB in the present study, the Hayling Test did not, and the reverse pattern was observed with respect to the age–Eyes Test association. Importantly, inhibitory mechanisms have been attributed with three general functions: restraint of dominant, but inappropriate, responses; prevention of access into working memory; and the deletion of information already in working memory (Hasher, Zacks, & May, 1999). Thus, different measures of inhibition are likely to vary in their relative demands on these aspects of inhibitory control (see Kramer et al., 1994), and they may therefore relate differentially to capacity for social reasoning.
Indeed, one possibility is that the Hayling Test may impose greater demands than the Stroop task on the prevention aspect of inhibitory control. This is because one of the strategies that a person may use to complete the Hayling Test is to preselect items that may serve as answers before the person hears the sentence that must be completed with an unconnected word. Participants who engage in this strategy may therefore successfully complete the task by reducing access to working memory of the sentence to be completed. This same strategy cannot be applied to the Stroop task as responses are predetermined within the test. Similarly, it may be that Eyes Test performance imposes greater demands on the prevention aspect of inhibition relative to high-inhibition FB performance. However, future research is needed to delineate the degree to which the access, restraint, and deletion functions of inhibitory control are implicated in ToM, and in socioemotional functioning more broadly.
There is increasing emphasis in psychology to regard cognitive functioning as being intertwined with social, motivational, and emotional processes. This is exemplified not only in the ToM literature (e.g., Carlson et al., 2004; German & Hehman, 2006; Henry, Phillips, Crawford, Iatswaart, & Summers, 2006) but also in other areas such as stereotyping. For example, von Hippel, Silver, and Lynch (2000) found that an increased propensity for older adults to rely on stereotypes was mediated by cognitive disinhibition. Thus, the current findings extend these researchers' suggestion that perceived age-related deficits in social reasoning may be due to unintentional rather than intentional processes; the findings also affirm the importance of further research that is focused on testing the potential interconnections between social and cognitive functioning. In addition, it has been argued that both reduced social collaboration and age-related declines in intellectual functioning undermine the acquisition of wisdom-related knowledge (Baltes & Staudinger, 2000). Thus, the present findings may have implications for the acquisition of wisdom in older adulthood.
In conclusion, empathy is an essential prerequisite for the development and maintenance of close interpersonal relationships (Fodor, 1987). Furthermore, loneliness and social isolation are known to have negative consequences for physical and mental well-being (House, Landis, & Umberson, 1988), particularly among older adults for whom reduced social participation has been linked to increased mortality (Fry & Debats, 2006). This highlights the importance of the present results, which not only provide further evidence that late adulthood is associated with increased difficulty seeing things from another person's point of view, but also identify social and nonsocial forms of inhibitory control as important mediators of this association. The inhibitory framework provided here has implications for understanding previously reported discrepancies in the age–ToM literature. That is, different perspective-taking, ToM, or cognitive empathy tasks may vary in the extent to which they impose demands on inhibitory mechanisms, and this has consequences for the relative performances of older and younger adults on these tasks.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication April 13, 2007. Accepted for publication December 19, 2007.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
