مکانیزم مجزا برای تاثیر حواس پرتی و وقفه در کار حافظه در سالخوردگی

Abstract Interference is known to negatively impact the ability to maintain information in working memory (WM), an effect that is exacerbated with aging. Here, we explore how distinct sources of interference, i.e., distraction (stimuli to-be-ignored) and interruption (stimuli requiring attention), differentially influence WM in younger and older adults. EEG was recorded while participants engaged in three versions of a delayed-recognition task: no interference, a distracting stimulus, and an interrupting stimulus presented during WM maintenance. Behaviorally, both types of interference negatively impacted WM accuracy in older adults significantly more than younger adults (with a larger deficit for interruptions). N170 latency measures revealed that the degree of processing both distractors and interruptors predicted WM accuracy in both populations. However, while WM impairments could be explained by excessive attention to distractors by older adults (a suppression deficit), impairment induced by interruption were not clearly mediated by age-related increases in attention to interruptors. These results suggest that distinct underlying mechanisms mediate the impact of different types of external interference on WM in normal aging.

1. Introduction Working memory (WM) involves the ability to store and manipulate information in mind over brief periods of time (Baddeley, 2003).1 WM involves multiple cognitive subsystems responsible for functions such as storage, rehearsal and executive functions (Miyake and Shah, 1999). While the rehearsal of information is a fundamental aspect of WM, executive control processes are necessary for optimal performance. This includes the ability to inhibit interference from information that intervenes during the period of memory maintenance (Sakai et al., 2002). Such interference has a negative impact on WM performance, likely due to disruption of active top-down control processes required to maintain relevant information (Baddeley, 1986, Sakai, 2003, Sakai and Passingham, 2004, Sreenivasan and Jha, 2007 and Yoon et al., 2006), as well as bottom-up disruption of stimulus representations in sensory cortices (Miller et al., 1996). Here we present a framework in which interference of WM maintenance may be caused by both internal (i.e., intrusions and diversions- internally generated thoughts/images (Forster and Lavie, 2009)) and external factors (i.e., distraction and interruption (Clapp et al., 2009)) (Fig. 1); the latter of which is the focus of the current study. Distraction involves encountered stimuli that are irrelevant and intended to be ignored (e.g., radio playing while attempting to rehearse a phone number). This filtering of irrelevant sensory input is thought to be dependent on top-down suppression signals from the prefrontal cortex (PFC) ( Chao and Knight, 1995 and Chao and Knight, 1998). Interruption by external interference involves intervening stimuli that are purposefully attended to as an aspect of a secondary task (e.g., a phone call while holding something in mind). An interruption requires a reallocation of cognitive resources, as well as processes involved in reactivating the disrupted representation afterwards, which is reliant on medial temporal lobe structures and the PFC ( Sakai and Passingham, 2004 and Sakai et al., 2002). For conceptual purposes, WM maintenance in the face of interruptions can be placed under the broader category of multitasking, which involves concurrently executed goal-directed operations ( Salvucci and Taatgen, 2008 and Salvucci et al., 2009). Recent network characterization with fMRI functional connectivity analysis revealed distinct mechanisms of influence by these two types of interference on WM maintenance ( Clapp et al., 2009). Encoded information is retained throughout a maintenance period even in the presence of distraction via connectivity between the PFC and sensory cortex, while interruption results in a disruption of this connectivity, and reactivation in the post-interruption maintenance period. Given such distinct mechanisms, the impact of distraction and interruption on WM in older adults, as well as the underlying etiology of a negative influence by these two types of interference, may be different. Interference conceptual framework. Fig. 1. Interference conceptual framework. Figure options It has been well demonstrated that WM performance declines with age (Dobbs and Rule, 1989, Foos and Wright, 1992 and Salthouse et al., 1991). However, it is believed that rote maintenance abilities are relatively spared in healthy aging (Craik and Jennings, 1992), and WM deficits are largely observed when executive processes are taxed (e.g., in the setting of interference or high memory load) (Gazzaley et al., 2007). Previous research has demonstrated that older adults have significant impairment in WM performance when they encounter interference, beyond that experienced by younger adults, (Gazzaley et al., 2008, Gazzaley et al., 2005b and Reuter-Lorenz and Sylvester, 2005). Internal interference, or intrusions, have been reported to disrupt an older adults’ ability to maintain information, as in studies of proactive interference (Emery et al., 2008 and Lustig et al., 2001) and internally generated thoughts (Borella et al., 2007). External interference by distraction disrupts WM performance in older individuals (Hasher et al., 1999) and is attributed to a deficit in top-down suppression of irrelevant information early in the visual processing stream (Gazzaley et al., 2008 and Gazzaley et al., 2005b). To our knowledge, no previous studies have directly addressed the impact of interruptions on WM in an older population. However, it has been shown that older adults are more disadvantaged than younger individuals when they divide their attention (Craik and Salthouse, 2000, Crossley and Hiscock, 1992, Kramer et al., 1995, Kramer and Larish, 1996, McDowd and Craik, 1988, Park et al., 1989 and Tsang and Shaner, 1998). These differences persist even when controlled for age-related performance decrements on a single task (Crossley and Hiscock, 1992). Likewise, the ability to multitask diminishes in older adults, as assessed by driving simulations (Chaparro et al., 2005 and Ponds et al., 1988), task management tests (Craik and Bialystok, 2006) and gait/posture experiments (Doumas et al., 2008 and Faulkner et al., 2007). The goal of this study was to explore the influence of these different types of external interference on WM in normal aging. To accomplish this, as well as to investigate the neural basis of any age-related behavioral effects, electroencephalography (EEG) was used to record neural activity as participants engaged in a cognitive paradigm assessing WM in the setting of distraction and interruption. Interference was introduced during the maintenance period of a simple delayed-recognition task. We performed this experiment in a group of healthy older participants and compared the data to those obtained from a population of younger adults who recently participated in the same experiment (Clapp et al., 2009). The inclusion of both types of external interference in the same experiment allows us to directly compare the consequences of age-related alterations in the suppression of irrelevant information and multitasking on WM performance. Neural analysis focused on early ERP measures associated with visual stimulus representation and attentional control

3. Results Results from the younger participants have been previously published (Clapp et al., 2009) and will be reported here for comparison purposes with data from the older participants. 3.1. Behavioral data ANOVA of WM accuracy was performed with the three WM tasks (NI, DS, IS) as the within-participant factor and age (young, older) as the across-participant factor. This analysis revealed a main effect of task (F(2,78) = 51.74, p < .001), such that participants performed with high accuracy when no interference was present, exhibited a significant reduction in accuracy with the presence of a distractor (NI vs. DS; p < .05) and a further decline in performance in the setting of an interruptor (NI vs. IS, p < .05, IS vs. DS, p < .05) ( Fig. 3A). Analysis also revealed a main effect of age (F(1,39) = 25.40, p < .001), such that older participants performed the WM tasks with lower accuracy. In addition, there was a significant task by age interaction (F(2,78) = 4.48, p < .05). Within-age group comparisons revealed that for both younger and older adults there was a WM accuracy decrement with the presence of a distractor and an interruptor (all comparisons; p < .05, see Table 1 for means and standard errors). Between-group comparisons showed that older participants performed worse than younger participants on all three WM tasks (all p < 0.05). Importantly, there was a disproportionately greater impact of both the distractor and interruptor on WM accuracy in the older group (NI–DS: younger vs. older, p < .05. NI–IS: younger vs. older, p < .05) ( Fig. 3B). Furthermore, there was a disproportionately greater impact of the interruptor relative to the distractor in older, compared to younger participants (IS–DS: younger vs. older, p < .05) ( Fig. 3B). All participants performed at or above 93% in the gender/age discrimination during the interruptor task. Working memory accuracy and impact of interference. (A) Participants performed ... Fig. 3. Working memory accuracy and impact of interference. (A) Participants performed best in the no interference task (NI), followed by the distractor task (DS), and then the interruptor task (IS) (all comparisons are significantly different, p < .05). The older participants performed with a lower WM accuracy on all tasks. (B) Older participants have a greater impact on WM performance by both the distractor (DS) and the interruptor (IS), when corrected by their performance without interference (p < .05). Note: Asterisks represent significant differences between age groups. Figure options Table 1. Comparisons of younger and older participants’ behavioral and neural data. Standard errors are presented in parentheses. Younger Older Behavioral data Accuracy Interrupting stimulus 89% (.007) 81% (.02) Distracting stimulus 93% (.01) 86% (.01) No interference 96% (.006) 92% (.009) Reaction time Interrupting stimulus 768 (28) 1200 (56) No interference 675 (23) 1110 (62) Neural data Encode P100 amplitude Interrupting stimulus 6.8 (0.6) 7.1 (0.6) Distracting stimulus 6.5 (0.6) 6.7 (0.52) No interference 6.3 (0.54) 7.2 (0.59) Passive view 6.7 (0.58) 6.6 (0.6) P100 latency Interrupting stimulus 107 (2.1) 117 (4) Distracting stimulus 108 (2.1) 116 (3.8) No interference 109 (2.2) 119 (3.9) Passive view 109 (2.2) 118 (3.7) N170 amplitude Interrupting stimulus −8.7 (0.7) −7.6 (0.83) Distracting stimulus −9.4 (0.96) −8.3 (0.76) No interference −9.6 (1.0) −7.6 (0.77) Passive view −9.9 (1.0) −7.6 (0.94) N170 latency Interrupting stimulus 173 (4.2) 189 (4.3) Distracting stimulus 173 (4.7) 189 (3.9) No interference 176 (4.4) 192 (4.2) Passive view 182 (4.1) 196 (4.1) Interference P100 amplitude Interrupting stimulus 6.9 (0.56) 6.7 (0.55) Distracting stimulus 6.2 (0.52) 6.3 (0.49) Passive view 6.8 (0.58) 6.3 (0.55) P100 latency Interrupting stimulus 106 (2.5) 120 (3.4) Distracting stimulus 109 (2.8) 122 (3.5) Passive view 108 (2.6) 122 (3.3) Passive view −8.9 (0.73) −8.60 (0.83) N170 amplitude Interrupting stimulus −7.7 (0.72) −8.2 (0.67) Distracting stimulus −7.8 (0.71) −7.73 (0.7) Passive view −8.9 (0.73) −8.60 (0.83) N170 latency Interrupting stimulus 173 (4.3) 189 (4.1) Distracting stimulus 178 (4.5) 190 (3.7) Passive view 176 (4.4) 193 (4.3) Table options The same ANOVA of WM reaction time, revealed a main effect of task (F(2,78) = 23.061, p < .001), such that participants performed with the fastest reaction times when no interference was present, exhibited a significant slowing in reaction time with the presence of a distractor, and further slowing in the setting of an interruptor (see Table 1). Analysis also revealed a main effect of age (F(1,39) = 62.10, p < .001), such that older participants performed the WM tasks with slower reaction times. There was no significant task by age interaction (F(2,78) < 1, p < .05). Analysis of the post-experiment long-term memory recognition measures involved first correcting each participant's LTM rating for each stimulus type with their LTM rating for novel stimuli, thus equating for response bias differences. ANOVA was then performed on these LTM indices with stimulus (IS-Cue, IS-Interruptor, DS-Cue, DS-Distractor, NI-Cue, PV-Cue, PV-intervening stimulus) and age (young, older) as factors. Analysis showed a main effect of stimulus (F(6,234) = 2.97, p < .05), no main effect of age (F(1,39) < 1), but a significant stimulus by age interaction (F(6,234) = 2.54, p < .05). The absence of a main effect of age revealed that both age groups remembered the same amount of information over the long-term from the experiment. To explore the age by stimulus interaction further, apriori t-tests revealed that older participants remembered the interfering stimuli from the IS and DS tasks as well as the cue stimuli; whereas the younger participants remembered the cue stimuli significantly better than the interfering stimuli (see Table 1 and Fig. 4). Both age groups remembered the cue stimuli better than novel stimuli (p < .05), but only the older group remembered the interfering stimuli (from IS and DS) better than the novel stimuli (p < .05) ( Fig. 4). Comparisons between age groups reveal that older participants remembered the interfering stimuli better over the long-term than the younger participants (p < .01). When the data was divided by interfering stimulus type, it revealed that older participants significantly remembered both distractors and interruptors better than the younger participants (IS-Interruptor: younger (M: −0.04, SE: 0.04) vs. older (M: −0.09, SE: 0.04), p < .05, DS-Distractor: younger (M: −0.03, SE: 0.045) vs. older (M: 0.088, SE: 0.03), p = .05) ( Fig. 4). There was no significant difference between age groups for memory of the cue stimuli. Incidental long-term memory recognition. Older participants recognized ... Fig. 4. Incidental long-term memory recognition. Older participants recognized interfering stimuli presented in the experiment more than younger adults and remembered them as well as the cues stimuli. Note: Single asterisks represent significant differences from 0, and double asterisks represent significant differences between ages/conditions. Figure options Importantly, in regards to the Passive View task, the cue and intervening stimuli presented in PV were less well remembered in the long-term compared to the cue stimuli from the WM tasks in both age groups. Furthermore, in both age groups, PV stimuli were not remembered differently than the novel stimuli. This result suggested that minimal attention was directed to these stimuli, as instructed. There were no significant correlations between neuropsychological measures and either neural or task performance measures. 3.2. Neural data Two posterior ERP measures have previously been shown to be modulated by attention and associated with visual processing: The P100 (50–150 ms) (Gomez Gonzalez et al., 1994) and N170 (120–220 ms) (Gazzaley et al., 2005a and Gomez Gonzalez et al., 1994). All data presented here are for ERPs time-locked to the onset of either cue or interfering stimuli. Analyses, when appropriate, focused on the modulation indices of enhancement and suppression (described in the methods). 3.2.1. Cue stimuli Analysis of the P100 amplitude in response to cue stimuli showed no effect of task (F(3,117) = 1.42, p > .05), age (F(1,39) < 1, p > .05) or interaction (F(3,117) = 2.007, p > .05). Analysis of the P100 latency showed no effect of task (F(3,117) = 1.34, p > .05), but a significant effect of age (F(1,39) = 4.04, p = .05), such that older participants had later P100 latencies than younger participants. No age by task interaction was observed (F(3,117) < 1, p > .05). Analysis of the N170 amplitude revealed no effect of task (F(3,117) = 2.172, p > .05), age (F(1,39) = 2.022, p > .05) or interaction (F(3,117) = 2.215, p > .05). Analysis of the N170 latency showed a significant effect of task (F(3,117) = 25.257, p < .05), such that all WM tasks (NI, DS, IS) had earlier N170 latencies compared to the PV task, and thus significant enhancement for the encoded stimuli (all p < .05), and a main effect of age (F(1,39) = 6.317, p < .05), which revealed that older participants had later N170 latencies than younger participants (see Table 1). There was no interaction between task and age (F(3,117) = 1.46, p > .05). To summarize, analyses of the cue period revealed a clear marker of attentional allocation (i.e., N170 latency), and an age-related slowing of neural processing (i.e., later P1 and N1 peaks in older adults), but none of the comparisons showed a significant age by task interaction. 3.2.2. Interfering stimuli This analysis focused on the same ERP markers as described above, but for the intervening face stimuli presented within the delay period of the IS, DS and PV tasks. ANOVA of the P100 amplitude showed a main effect of task (F(2,78) = 4.61, p < .05), such that the largest P100 was observed for the IS-Interruptor, but there was no effect of age (F(1,39) > .05) or interaction between age and task (F(2,78) = 2.60, p > .05). P100 latency showed a main effect of task (F(2,78) = 5.129, p < .05), such that responses to IS were faster than PV and responses to DS. A main effect of age (F(1,39 = 11.59, p < .05) was also shown, once again revealing slowing of P100 latencies with age. There was no significant interaction between age and task (F(2,78) > .05). N170 amplitude analysis revealed only a significant effect of task (F(2,78) = (11.391, p < .05), such that responses to IS were highest, followed by PV and DS. Neither the effect of age (F(1,39 < 1) nor the interaction between age and task were significant (F(2,78) = 2.19, p > .05). Analysis of N170 latency revealed a main effect of task (F(2,78) = (19.67, p < .05), a main effect of age (F(1,39) = 6.41, p < .05), and a significant task × age interaction (F(2,78) = 4.84, p < .05). This measure has been revealed to reflect attention, as well as age-related changes in visual processing in previous studies ( Gazzaley et al., 2008 and Gazzaley et al., 2005a). Post hoc t-tests revealed that the N170 latency for IS was earlier than for PV and DS (main effect of task), and a significantly later N170 latency was observed in older adults (main effect of age), consistent with reports in the literature of generalized slowing of processing speed with aging ( Pfefferbaum et al., 1984) (see Fig. 5 and Table 1). Grand-averaged event-related potentials to intervening stimuli. (A) GAV ERPs ... Fig. 5. Grand-averaged event-related potentials to intervening stimuli. (A) GAV ERPs from younger participants for Distractor stimuli (DS), Interruptor Stimuli (IS) and passively viewed intervening stimuli (PV). (B) ERPs from older participants. (C) N170 latency measures to intervening stimuli. Both younger and older participants significantly enhanced the interruptors (earlier peaks compared to passively viewed intervening stimuli). Unlike younger adults, older participants also showed enhancement for the distracting stimuli. (D) Comparisons of N170 latency attentional modulation indices between age groups. Older participants allocated more attention towards distractors than younger participants (suppression index), but did not differ in attention allocated towards interruptors (enhancement index). Figure options Analysis directed at interpreting the interaction included both within-group and between-group comparisons. For within-group comparisons, both younger and older participants displayed earlier latencies for the interruptor in the IS task than the intervening stimuli in the PV task, thus both displaying significant enhancement (young: IS-Interruptor vs. PV-intervening stimulus, p < .05. older: Interruptor vs. PV-intervening stimulus, p < .05.) ( Fig. 5A–C). Earlier N170 latencies for relevant stimuli compared to passively viewed stimuli has been previously revealed in younger adults ( Clapp et al., 2009 and Gazzaley et al., 2005a) and is interpreted to reflect greater attention to the stimulus as mediated by more rapid synchronization of cortical areas involved in stimulus representation. Neither age group displayed significant suppression of the N170 latency for the distractors (i.e., later latency for DS than PV). Conversely, the older participants actually exhibited significant enhancement of the distractor in the DS task (p < .05) (i.e., earlier latency for the distractor than the passively viewed stimulus), while younger participants did not show this pattern. Between-group comparisons revealed a significant deficit in the N170 latency suppression index for distractors by older participants (p < .01), but no significant age-related difference in the enhancement index for the interruptors (p = 0.658) ( Fig. 5D). 3.3. Comparisons between cue and interfering stimuli N170 latency enhancement was compared between cue and interruptor stimuli from the IS task to determine if participants directed different degrees of attention to these two relevant stimuli. The N170 latency was utilized as the measure of comparison because it was the only early measure to reveal significant modulation for both cue and interfering stimuli. This analysis revealed that younger participants exhibited significantly greater enhancement for cues than interruptors (p < .05), while the older participants did not differ in their enhancement for these stimuli (p > 0.05). A between-group analysis reveals that younger and older participants did not differ in the amount that they enhance cue or interruptor stimuli (p > 0.05). 3.4. Neural–behavioral correlations To evaluate if the magnitude of activity modulation to the cue or interfering stimuli predict WM performance, across-participant regression analyses were performed for N170 latency modulation indices (i.e., enhancement and suppression) and WM accuracy. Modulation indices were utilized because responses relative to passively viewed stimuli equate for each individual's overall ERP differences. Analyses of the interfering stimuli in older participants paralleled the results obtained from the same analysis in the younger participants (Clapp et al., 2009): (1) There was a positive correlation for both age groups between the suppression of distractors in the DS task and WM accuracy (young: R = 0.49, p < 0.05; older: R = 0.50, p < 0.05), such that those individuals who neurally suppressed distracting information the most, showed superior WM performance ( Fig. 6A). (2) There was a negative correlation between the amount of enhancement to interruptors in the IS task and WM accuracy (young: R = −0.77, p < 0.05; older: R = −0.45, p < 0.05), which revealed that those participants, both younger and older who enhanced the interruptor most exhibited the worst WM performance ( Fig. 6B). (3) In addition, a negative correlation existed between indices of enhancement of interruptors and suppression of distractors using the N170 latency indices in both younger and older populations (young: R = −0.7, p < 0.05; older: R = −0.55, p < 0.05), revealing that those individuals who enhance the interruptor the most, also suppress distractors the least. Neurobehavioral correlations of the N170 latency attentional modulation indices. ... Fig. 6. Neurobehavioral correlations of the N170 latency attentional modulation indices. In both (A) younger and (B) older age groups, participants that enhance the interruptor more (upper panels) perform worse on the WM task. Likewise, participants that suppress the distractor more (upper panels) perform better on the WM task. Note: More positive value on the x-axis indicate greater enhancement above baseline or greater suppression below baseline. Figure options 3.5. Practice effects Behaviorally, performance improvement in the DS task across blocks was observed for both age groups, reflected as a decreased RT in the younger participants (p < .05, see Table 2 for means and standard errors), and increased WM accuracy in the older participants (p < .05). In the NI and IS tasks, a non-significant increase in WM accuracy was observed in both younger and older adults. Table 2. Practice effects in younger and older participants. Standard errors are presented in parentheses. Practice effects Younger Older Block 1 Block 2 Block 1 Block 2 Behavioral findings Accuracy IS 88% (2.0) 91% (1.5) IS 79% (2.1) 81% (1.8) DS 92% (1.5) 94% (1.6) DS 84% (1.6) 88% (1.5) NI 95% (1.4) 96% (1.4) NI 90% (1.6) 91% (1.1) Reaction time IS 824 (34) 798 (28) IS 1255 (54) 1189 (48) DS 717 (22) 651 (22) DS 1133 (55) 1151 (54) NI 678 (24) 650 (27) NI 1159 (57) 1089 (50) Neural Encode IS 173 (4.3) 172 (4.1) IS 187 (3.2) 188 (4.1) DS 178 (4.5) 180 (3.7) DS 189 (3.5) 191 (4.0) NI 176 (4.4) 173 (3.3) NI 191 (3.0) 192 (5.1) Interference IS 824 (34) 798 (28) IS 1255 (54) 1189 (48) DS 717 (22) 651 (22) DS 1133 (55) 1151 (54) NI 678 (24) 650 (27) NI 1159 (57) 1089 (50) Table options An ANOVA was used to evaluate practice effects for the neural response to cue and interfering stimuli, using task (DS, IS, NI, PV), block (first, second) and age (young, older) as factors. We again focused on the N170 latency, as it showed significant attentional modulation in both populations. For the cue stimuli there was a main effect of task (F(3,117) = 15.15, p < .05) and age (F(1,39) = 11.62, p < .05), but no effect of block (F(1,117) < 1), no task by age interaction (F(3,117) < 1), no block by task interaction (F(3,117) = 1.03, p > .05), and no block by task by age interaction (F(3,117) = 1.19, p > .05). For the interfering stimuli there was a main effect of task (F(2,78) = 7.07, p < .05), block (F(1,78) = 8.17, p < .05) and age (F(1,39) = 5.29, p < .05), a block by task interaction (F(2,117) = 7.22, p > .05), but no task by age interaction (F(2,117) < 1) or block by task by age interaction (F(2,117) = 1.31, p > .05). t-tests revealed that the latency of the N170 to distractors (DS task) were significantly later in the second block (younger and older, p < .05), but there was no across block change for the interruptors (IS task)—see Table 2 for values. Comparisons of modulation indices, showed that suppression of distractors increased in the second block (younger and older, both p < .05), while enhancement of interruptors remained the same across blocks. Taken together with the behavioral results, this suggests the possibility that performance improvements in the DS task may have been driven by changes in the response to distractors by both younger and older adults, as has been determined in a recent study of younger adults using the same design, but with lower level stimuli (moving flow fields of dots) ( Berry et al., 2009).