209 SELF-AWARENESS AND AGING 3 Main Text: 5507 MAIN TEXT: We are all prone to occasional lapses or errors as we engage in our daily activities. Although the errors themselves typically have unfavorable consequences, they can also play a critical adaptive role by signaling to us that current performance levels are not sufficient to attain our goals, and by allowing us to establish accurate impressions of our own abilities. The capacity to monitor performance errors is particularly important from a clinical perspective, as compromised self-awareness (SA) of a deficit will necessarily impede the patient in making efforts to recover from it, or implement compensatory strategies. Indeed, numerous studies have documented associations between poor SA of deficits and a range of negative outcomes, including poor motivation for treatment (Fleming, Strong & Ashton, 1996; Malec & Moessner, 2001), increased care-giver burden (Seltzer, Vasterling, Yoder, & Thompson, 1997) and poor general prognosis (David, 1992; McEvoy, Apperson, Appelbaum, & Ortlip, 1989). The natural aging process is known to have a deleterious effect on a wide range of cognitive functions (Grady, 2012; Heddin & Gabrieli, 2004; McAvinue, Habekost, Johnson, et al., 2012; Salthouse, 1996), rendering older adults more prone to erroneous behavior (e.g. Burke & Shafto, 2004; Gold, Powell, Xuan, Jicha & Smith, 2010; Young & Bunce, 2011), yet very little research has examined SA of deficits in healthy older adults. Functional imaging work has suggested that the neural substrates of SA reside across a distributed network of brain regions (Pia, Neppi-Modona, Ricci, et al. 2004; Prigatano & Schacter, 1991; Rosen, Alcantar, Rothlind et al., 2010), but the robust association between compromised prefrontal cortex (PFC) function and SA deficits across several different clinical populations, including traumatic brain injury SELF-AWARENESS AND AGING 4 (O’Keeffe, Dockree, Moloney, Carton, & Robertson, 2007), schizophrenia (David, Bedford, Wiffen & Gilleen, 2012), substance abuse (Hester, Nestor, & Garavan, 2009), Alzheimer’s Disease (Starkstein, Vázquez, Migliorelli, et al., 1995), attentiondeficit/hyperactivity disorder (O’Connell, Bellgrove, Dockree, et al., 2009) and focal frontal lesions (Hoerold, Pender & Robertson, 2013), indicates that the PFC is a particularly important component of the SA network. Although the neuropsychological underpinnings of SA have yet to be fully established (Prigatano, 2005) reduced SA has frequently been linked to memory impairment (e.g. Agnew & Morris, 1998; Starkstein et al., 1995; Noe et al., 2005), and findings within our own laboratory have demonstrated a close relationship between SA and sustained attention (for an overview, see Robertson (2010)). Given that the PFC is particularly vulnerable to the effects of aging (e.g. Heddin & Gabrieli, 2004), and that memory and attentional capacities are known to decline with increasing age (see Balota, Dolan, & Duchek, 2000 for a review), there is basis for hypothesizing that the capacity for SA may be reduced in older adults. Findings from the field of electrophysiology have provided some valuable information on how the aging brain responds to errors. Of particular importance has been the study of two event-related brain potentials (ERP) that occur when people make errors on laboratory tasks: the error negativity (Ne; Falkenstein, Hohnsbein, Hoormann, & Blanke, 1990) and the error positivity (Pe; Falkenstein, Hohnsbein, Hoormann, & Blanke, 1991). The precise functions of the Ne and Pe are still under debate, but it is generally assumed that the Ne reflects an early, rapid, and possibly preconscious, detection mechanism that is sensitive to response conflict (van Veen & Carter, 2002) or changes in reward probability (Holroyd & Coles, 2002). The Pe, on the other hand, has been linked to conscious error awareness, owing to the consistent SELF-AWARENESS AND AGING 5 observation that it is only present on trials where participants are aware of their errors (Endrass, Franke, & Kathmann, 2005; Murphy, Robertson, Allen, Hester & O’Connell, 2012; Niewenhuis, Ridderinkhof, Blom, Band, & Kok, 2001; O’Connell, Dockree, Bellgrove et al., 2007). A number of studies have found that the amplitudes of both the Ne (Band & Kok, 2000; Falkenstein, Hoormann, Christ, & Hohnsbein, 2000; Mathalon et al., 2003; Mathewson, Dywan & Segalowitz, 2005) and the Pe (Band & Kok, 2000; Falkenstein, Hoormann, Christ, & Hohnsbein, 2000; Leuthold & Sommer, 1999; Mathewson et al., 2005) are reduced in older adults relative to young controls. However, participants did not overtly signal their errors in these studies and therefore it cannot be determined to what extent attenuation of the trial-averaged Pe results from reduced error awareness or from differences in the processing of consciously detected errors. Outside of the field of electrophysiology, a small number of studies have suggested that older adults demonstrate a diminished ability to monitor and appraise performance (Bruce, Coyne, & Botwinick, 1982; Graham, Kunik, Doody, & Snow, 2005; Rabbitt, 1990; Suchy, Kraybill, Frnachow, et al., 2011), but others have provided some evidence to the contrary (Clare, Whitaker, & Nelis, 2010; Lovelace & Marsh, 1985; Rabbitt, 2002). Overall, the research on SA in healthy aging is not conclusive, and of particular import, these studies have not measured SA across a range of cognitive and behavioral domains, which is important given the potential domain specificity of SA deficits (e.g. Hart et al., 2005; Hart, Sherer, Whyte et al., 2004; Prigatano & Altman, 1990). In the present study we aimed to address this gap in the literature by employing the first multi-domain assessment of SA in healthy older adults. SELF-AWARENESS AND AGING 6 The most common method for measuring SA in patient populations is to examine the discrepancy between self-reports and informant-reports on questionnaire measures of daily functioning, with the premise that a discrepancy in the direction of the informant reporting more difficulties indicates impaired SA (Fleming, Strong & Ashton, 1996; Hart et al., 2004). We examined SA in terms of attentional control, memory functioning and socio-emotional functioning, respectively, using this questionnaire discrepancy score method. We also administered a computerized measure of SA that required participants to overtly signal their errors (i.e. demonstrate online error awareness) during a neuropsychological task. A number of authors have argued that online error awareness enables recognition of difficulties as they occur, and may therefore contribute to broader aspects of SA in daily life (Jenkinson, Edelstyn, Drakeford, & Ellis, 2009; Larson & Perlstein, 2009; Ownsworth & Fleming, 2005; Robertson, 2010). However, such a relationship has yet to be established empirically, and was accordingly identified as an important question for the current study. A battery of neuropsychological tests were also administered to obtain cognitive profiles of the participants and to examine the relationship between SA and other cognitive domains. Given that SA is linked to PFC function, as well as cognitive capacities such as attentional control and memory, all of which are known to deteriorate with increasing age, we hypothesised that older adults would have diminished SA relative to young adults. We also predicted that online error awareness would be associated with questionnaire measures of SA, and that SA would correlate positively with sustained attention and memory capacities. SELF-AWARENESS AND AGING 7 METHOD Participants Fifty-one older adults and 47 young adults took part in the study. Four older adults were excluded because their Mini-Mental State Examination (MMSE, (Folstein, Folstein & Hugh, 1975) score indicated possible cognitive impairment (<24 ). Two older and two younger adults were also excluded due to poor accuracy on the Error Awareness Task (<30% correctly withheld no-go trials). As a result, the final sample consisted of 45 younger adults (31 female) with a mean age of 22.7 years (SD 4.9, range 18-34) and 45 older adults (29 female) with a mean age of 76.2 years (SD 7.1, range 66-90). Exclusion criteria were visual impairment, history of psychiatric illness, neurological insult, drug or alcohol abuse, and/or reporting current use of antipsychotic or antidepressant medications. The most common illnesses for which older adults were taking medication for were hypertension (n=10), osteoporosis (n= 5), arthritis (n= 5), and hypothyroidism (n= 4). All participants were asked to refrain from consuming caffeine on the day of testing. Procedures were approved by the ethical review board of the School of Psychology, Trinity College Dublin in accordance with the Declaration of Helsinki, and all participants provided informed consent. Background Measures A number of background neuropsychological tests and measures were administered to all participants. These included: The Mini-Mental State Examination (MMSE), the National Adult Reading Test (NART; Nelson, 1982; Nelson and Willison, 1991), Logical Memory 1 (immediate recall) subtest of the Wechsler SELF-AWARENESS AND AGING 8 Memory Scale (WMS-III; Wechsler, 1997), a test of verbal fluency (animal naming), the Sustained Attention to Response Task (SART; Robertson et al., 1997), and a twochoice reaction time task (CRT). Participants also completed the Hospital Anxiety and Depression Scale (HADS; Zigmond and Snaith, 1983) to assess symptoms of anxiety and depression. For both the SART and the CRT, stimuli were presented on a Dell Latitude Laptop using E-prime 2.0 software (Psychology Software Tools, Inc, PA, USA). For the SART, the numbers 1 to 9 appeared in a fixed sequence and participants were required to make a left button click for every number except for the number three. Participants completed two blocks of the task, each of which included 25 No-go targets (the number 3) and 200 Go trials (all other numbers). The CRT was included as a measure of speed of processing. Each trial of the CRT started with the participant holding down a white central button on a RB-530 response box (Cedrus, San Pedro, CA, USA) to trigger target onset. Participants were then required to make a speeded press of a ‘green’ button if the target ‘YES’ appeared on screen or a ‘red’ button if the target ‘NO’ appeared. They were instructed to return to the white button after target offset to trigger the next target. To guard against pre-emptive responding target offset could not be achieved if the white button was released before the target appeared. The task comprised 50 ‘NO’ trials and 50 ‘YES’ trials presented in random order. The interval between depression of the white trigger button and the target onset varied between 800ms and 1100 ms. The task was self-paced, but participants were instructed to respond quickly and accurately. Speed of processing was measured in milliseconds and was split into a ‘cognitive response’ measured as time from target onset to trigger offset (release of white key) and ‘motor response’ measured as time from trigger offset to response selection (depression of green key or red key). SELF-AWARENESS AND AGING 9 Awareness Measures The Error Awareness Task (EAT; Hester, Foxe, Molholm, Shpaner, & Garavan, 2005) was used as a measure of online error awareness, and questionnaire measures of daily functioning included the Cognitive Failures Questionnaire (CFQ; Broadbent, Cooper, Fitzgerald, & Parkes, 1982), the Socio-Emotional Questionnaire (SEQ; Bramham, Morris, Hornak, Bullock, & Polkey, 2009) and the Memory Awareness Rating Scale, Memory Functioning Scale (MFS; Clare et al., 2002). The Error Awareness Task The EAT is a Go/No-go response inhibition task in which participants are presented with a serial stream of single color words, with congruency between the semantic meaning of the word and its font color manipulated across trials. Participants were trained to respond with a single speeded left mouse button press in situations where the meaning of the word and the font color in which it was presented were congruent (Go trial) and to withhold this response when either of two different scenarios arose: (1) when the word presented on the current trial was the same as that presented on the preceding trial (Repeat No-go trial), and (2) when the meaning of the word and its font color did not match (Incongruent No-go trial). In the event of a commission error (failure to withhold to either of these No-go trials) participants were trained to signal their “awareness” by making a right mouse button press on the subsequent trial. In these instances they were not required to make their standard Gotrial response. The next standard Go-trial after an error was thus rendered irrelevant which guarded against the possibility that some errors may fail to reach awareness because ongoing processing has been interrupted by the onset of another stimulus (Rabbitt, 2002). SELF-AWARENESS AND AGING 10 In addition, due to concerns that group differences in online error awareness on this task might arise purely from group differences in the number of errors made, we integrated a feature that adaptively modified the difficulty of the task. This entailed checking the participants’ accuracy over consecutive periods of 40 trials and adapting the stimulus duration accordingly. The first 40 stimuli of the task were always presented for 750 ms and were succeeded by an inter-stimulus interval (ISI) of 750 ms. The stimulus duration subsequently remained at 750 ms as long as accuracy on the previous 40 trials was between 50% and 60%. However, if accuracy exceeded 60% the stimulus duration and ISI were set to 500ms and 1000ms respectively for the subsequent 40 trials. If accuracy fell below 50%, the stimulus duration and ISI were set to 1000ms and 500 ms respectively. This evaluation and task adjustment occurred every 40 trials thereafter. All participants performed four blocks of the task, consisting of 225 word presentations, 200 of which were Go trials and 25 of which were No-go trials (12 Repeat No-gos and 13 Incongruent No-gos, or vice versa). The duration of each block was approximately 5.6 min. It was ensured that all participants were well-practiced and fully understood the requirements of the task before they began their first block. The Cognitive Failures Questionnaire The CFQ is a 25–item scale that includes statements relating to levels of attentional control in daily life. It has been employed in a broad range of clinical and non-clinical populations and has high construct validity (e.g. Larson et al., 1997; Wallace, Kass, & Stanny, 2002; Wallace & Vodanovich, 2003). The specificity of the CFQ as a measure of attentional control, rather than global cognitive function, is borne out by research indicating that the scale is not correlated with general SELF-AWARENESS AND AGING 11 intelligence but is robustly correlated with objective indices of attention (Manly, Robertson, Galloway & Hawkins, 1999; Robertson et al., 1997; Tipper & Baylis, 1987). Higher CFQ scores indicate poorer perceived attentional control. The Memory Functioning Scale The MFS is comprised of 13-items that ask about individuals ability to perform memory tasks in a range of everyday situations. The scale has been validated in healthy aging and early stage Alzheimer Disease (Clare, Whitaker & Nelis., 2010). Higher MFS scores indicates better perceived memory functioning. The Socio-Emotional Questionnaire The SEQ is a 30-item scale that includes statements relating to the recognition of basic emotions, empathy with the expression of these emotions, relationship skills and public behavior. The SEQ has demonstrated reliability and validity in brain injury patients (Bramham et al, 2009) and healthy adolescents (Wall, Williams, Morris & Bramham, 2011). Higher scores indicate poorer perceived socioemotional functioning. Each of the questionnaire measures of awareness was rated for identical items by participants and an informant. All informants were aged between 20 and 64 years (Mean 48.16, SD 10.59), had known the participant for 2 years or more, and had spent 6 hours or more with the participant in the 2 months preceding completion of the questionnaires. Discrepancy scores were calculated correcting for differences in direction of scoring. The difference between self-ratings and informant ratings were divided by the mean of the two sets of ratings to prevent scaling effects from SELF-AWARENESS AND AGING 12 distorting the measurement (Clare, Whitaker, & Nelis, 2010; Clare, Whitaker, Nelis, et al., 2011). Corrected discrepancy scores close to zero indicate good agreement between the participant and the informant. For all three measures, positive scores indicated that the informant reported more difficulties than the participant, and vice versa. Statistical Analysis All of the neuropsychological tests, as well as performance indices on the EAT were analyzed using one-way ANOVAs. Due to the fact that response times (RT) to Go-trials following a No-go trial were likely to be disrupted by errorsignaling responses, trials up to n+1 relative to the error-signaling response were excluded from the Go-trial RT analysis. For the questionnaire measures of SA, univariate ANCOVAs were performed on each of the corrected discrepancy scores with Age Group (two-level) as the between subjects factor. Significant group differences were found for speed of cognitive response, speed of motor response, anxiety and depression, therefore all of these variables were entered as covariates. Significant main effects (p<.05) were followed up with Bonferroni adjusted paired and independent samples t-tests to determine the origin of the effect of Age Group. All reported effects were significant regardless of whether the covariates were included or not. To examine, the interrelationships between the different domains of SA, and the extent to which the domains of SA related to the cognitive capacities for sustained attention, memory and verbal fluency, we conducted one-tailed Bonferroni adjusted partial correlations controlling for speed of cognitive response, speed of motor response, anxiety and depression. Again, all reported effects were significant regardless of whether the covariates were included or not. SELF-AWARENESS AND AGING 13 RESULTS The demographic and neuropsychological data for both groups are summarized in Table 1. The groups were successfully matched for sex (χ(1) = .20, p = .655) years of education (F(1,88) = .076, p = .783) and estimated IQ (F(1,88) = 3.63, p = .060). Young adults reported higher levels of anxiety (F(1,88) = 20.19, p < .001) and depression (F(1,88) = 11.40, p < .01) than older adults. Significant effects of Age Group were observed for all of the background neuropsychological tests. Older adults had significantly lower MMSE (F(1,88) = 23.36, p < .001), memory (F(1,88) = 25.69, p < .001), sustained attention (F(1,88) = 10.60, p < .01) and verbal fluency (F(1,88) = 38.43, p < .001) scores than young adults. Older adults also had a slower cognitive response (F(1,88) = 46.75, p < .001) and slower motor response (F(1,88) = 75. 80, p < .001), compared to young adults. Thus, although all participants were within the normal range for healthy older adults, older adults showed the expected age-related decline in cognitive function.