The differential contribution of executive functions to temporal generalisation, reproduction and verbal estimation

Highlights

• Temporal tasks differentially recruit executive resources.

• Updating, switching and access to semantic memory influence timing ability.

• Inhibition does not influence single-task timing performance.

Abstract

Evidence from dual-task studies suggests that executive resources are recruited during timing. However, there has been little exploration of whether executive recruitment is universal across temporal tasks, or whether different temporal tasks recruit different executive resources. The current study explored this further by examining how individual differences in updating, switching, inhibition and access affected performance on temporal generalisation, reproduction and verbal estimation tasks. It was found that temporal tasks differentially loaded onto different executive resources. Temporal generalisation performance was related to updating and access ability. Reproduction performance was related to updating, access and switching. Verbal estimation performance was only related to access. The results suggest that executive resources may be recruited when monitoring and maintaining multiple durations in memory at the same time, and when retrieving duration representations from long-term memory. The findings emphasise the need to consider timing behaviour as the product of a wide range of complex, integrated, cognitive systems, rather than as the output of a clock in isolation.

Introduction

Recent research has demonstrated that working memory, attention and executive functions are critical to temporal perception (Block et al., 1998, Brown, 2006, Brown, 2014, Brown et al., 2013, Droit-Volet and Zélanti, 2013, Fortin et al., 2010, Mioni et al., 2013, Ogden, Salominaite, Jones, Fisk and Montgomery, 2011, Perbal et al., 2002, Pouthas and Perbal, 2004, Rattat, 2010, Wearden et al., 1997, Zakay and Block, 2004, Zélanti and Droit-Volet, 2011, Zélanti and Droit-Volet, 2012). There remains, however, a lack of clarity regarding specifically which working memory and executive functions are involved in timing, and the extent to which their involvement is common to all temporal tasks. Consequently, this study aimed to explore how executive functions may be differentially involved in different timing tasks.

Research examining the involvement of executive resources in timing has generally used an interference paradigm in which performance on timing and non-timing tasks is compared under single and dual-task conditions (see Brown, 2006 for review). In a typical experiment a participant would complete a timing task, such as serial interval production, and an executive task, such as serial subtraction of sevens, alone and concurrently under dual-task conditions. Of key interest is bidirectional interference between the two tasks. Poorer performance on both tasks under dual than single-task conditions indicates that both tasks are competing for the same limited resources (Navon & Gopher, 1979). Such experiments have demonstrated bidirectional interference, confirming that timing is to some extent dependent on executive resources (Brown, 2006, Brown, 2014, Brown et al., 2013, Ogden, Salominaite, Jones, Fisk and Montgomery, 2011, Ogden, Wearden, Gallagher and Montgomery, 2011, Rattat, 2010).

Recent theoretical models of executive function suggest that the central executive is not a unified construct and is instead made up of different components (Fisk and Sharp, 2004, Miyake et al., 2000). Miyake et al. (2000) identified three component processes of the central executive; updating, inhibition and switching. Updating refers to an individual's ability to monitor incoming information and to update the contents of working memory accordingly. Inhibition refers to an individual's ability to inhibit a dominant or automatic response when it is inappropriate. Switching refers to an individual's ability to switch their attention between different tasks or different elements of the same task. Fisk and Sharp (2004) added a fourth component, access, which reflects the efficiency of access to semantic memory. The fractionation of the central executive into these component processes is supported by neuroimaging studies. For example, updating tasks activate the dorsolateral prefrontal cortex (DLPFC; Goldman-Rakic, 1996), inhibition is associated with prefrontal cortex activity (Casey et al., 1997, Kiefer et al., 1998), switching produces anterior cingulate (Posner & Raichle, 1994) and left frontal lobe activity (Rogers et al., 1998), and access to semantic memory is associated with left PFC activity (Kolb & Whishaw, 1985). In addition, the maintenance of working memory as a whole is related to the integrity of the PFC (Miyake et al., 2000). Recent research has sought to establish which of these component processes is recruited during timing (Brown et al., 2013, Ogden, Salominaite, Jones, Fisk and Montgomery, 2011), and the results are summarised below.

Brown and Frieh (2000) and later Ogden, Salominaite, et al. (2011) suggested that because updating requires the online maintenance, temporal tagging and replacing of information in working memory, updating resources are likely to be recruited during timing. This suggestion is generally supported (Brown and Frieh, 2000, Brown et al., 2013, Droit-Volet and Zélanti, 2013, Ogden, Salominaite, Jones, Fisk and Montgomery, 2011, Ogden, Wearden, Gallagher and Montgomery, 2011, Zélanti and Droit-Volet, 2011, Zélanti and Droit-Volet, 2012). Ogden, Salominaite, et al. (2011) explored how updating, inhibition, switching and access were involved in timing. A dual-task paradigm was employed in which participants completed serial production and executive tasks under single and dual-task conditions. Updating was assessed by Serial Subtraction of Sevens (SSS) in which participants have to serially deduct 7 from 385 or 368 (counterbalanced for single vs. dual-task), verbalising their response. Bidirectional interference was observed; when concurrently timing and performing SSS, temporal productions became longer and more variable and there were significantly fewer correct subtractions in the SSS task. Ogden, Salominaite, et al. (2011) therefore concluded that interval production requires updating resources.

Brown et al. (2013) and Brown and Frieh (2000) found partial evidence for bidirectional interference between timing and updating tasks. Brown et al. (2013) used a modified version of the Mental Counters task (Larson, Merritt, & Williams, 1988) in which participants had to monitor the presentation of a series of geometric shapes and report how frequently certain shapes occurred. The Mental Counters task and serial production of 5 s were performed under single and dual-task conditions. In addition, participants were instructed how to share their attention between the two tasks; with either 75%, 50% or 25% of attention being dedicated to the updating task. Relative to single-task conditions, dual-task timing became more variable and productions lengthened as decreasing attention was paid to time. Timing only impaired updating performance when attention to the updating task was low (25%), resulting in slower response times. Partial bidirectional interference was also reported between serial production and a running memory task in Brown and Frieh (2000). Timing only interfered with running memory performance when the running memory task was easy (recall the last 3 items presented) but not when it was difficult (recall the last 5 items presented). It should be noted however that performance was equally poor on the difficult task under single and dual-task conditions, perhaps reflecting a ceiling effect, and the importance of assessing simple span in such tasks prior to running span.

Brown et al. (2013) suggest that shifting resources are likely to be recruited during timing because our day to day interactions with the environment require people to shift between timing and non-timing tasks. Partial support for this suggestion is provided (Brown et al., 2013, Wearden et al., 2010, Zakay and Block, 2004). Brown et al. (2013) observed bidirectional interference between serial interval production and the Local–Global task (a switching task requiring shifting attention between attention to individual items of a stimulus or the stimulus as a whole). Interval productions became longer and more variable under dual-task conditions. Response times in the Local–Global task were also greater under dual-task conditions, indicating that switching resources are recruited during timing. Interestingly, Wearden et al. (2010) observed interference between task-switching and timing without using a dual-task paradigm. The experiment involved conditions in which participants only completed a timing task (estimation and production) and other conditions in which participants switched from an addition task to a timing task. When switching from addition to timing, duration estimations became shorter than in conditions where only timing was performed. Wearden et al. (2010) suggested that task switching may have reduced attention to time, resulting in shorter perceived durations.

Interference between switching and timing is not however a universal finding (Fortin et al., 2010, Ogden, Salominaite, Jones, Fisk and Montgomery, 2011). Ogden, Salominaite, et al. (2011) observed that relative to single-task conditions, interval production became more variable when performed concurrently with the plus–minus task which requires participants to switch their attention between adding and subtracting from a list of 2-digit numbers. Concurrent interval production did not, however, affect the switch-cost (the time difference between the switch vs. no switch conditions). Similarly, Fortin et al. (2010) reported four experiments requiring participants to produce intervals whilst performing a memory search task, or switching between performing a memory search and digit classification task. Switching did not affect interval production. Both Ogden, Salominaite, et al. (2011) and Fortin et al. (2010) therefore concluded that switching resources are not always recruited during timing.

The role of inhibition in timing is presently unclear. Studies in which timing and the Stroop task are performed concurrently consistently report that perceived time is shorter under dual than single-task conditions; however, examination of bidirectional interference is rare. Brown et al. (2013) explored bidirectional interference between concurrent performance of the Stroop and serial 5-second production. Bidirectional interference was observed: relative to single-task conditions, interval productions became longer and more variable under dual-task conditions, with the greatest effects observed when attention to time was low (25%). Concurrent timing also lengthened response times on the Stroop task with greater effects when less attention was paid to the Stroop task (25%). Bidirectional interference was also reported by Brown (2006) using Random Number Generation (RNG). Ogden, Salominaite, et al. (2011) however, did not observe bidirectional interference between timing and concurrent performance of Random Letter Generation (RLG). Whilst RLG made interval production more variable, timing did not influence RLG. The disparity between Ogden, Salominaite, et al. (2011) and Brown's (2006) findings can perhaps be explained by varying levels of task difficulty; Brown (2006) required participants to produce numbers at a faster rate (every 0.86/s) than Ogden, Salominaite, et al. (2011) who required participants to produce letters once every second. Indeed, the linear relationship between attention to the Stroop task and Stroop interference, reported in Brown et al. (2013), supports the suggestion that inhibitory resources are perhaps only required for timing when task demands are high.

The role of access to semantic memory in timing is rarely explored. Ogden, Salominaite, et al. (2011) provide the only exploration of bidirectional interference between timing and access. Participants completed serial interval production alone and concurrently with the Controlled Oral Word Association (COWA) task. The COWA task requires participants to retrieve as many words as possible beginning with a given letter (F, A, or S) in 1 min. Although bidirectional interference was not observed in this instance, Ogden, Salominaite, et al. (2011) acknowledge that access resources may be recruited on different types of temporal tasks. One potential role for access is in the retrieval of representations of duration from long-term memory. Durations used on more than one occasion (e.g. the standard in a temporal generalisation task) are thought to be stored in a long-term reference memory (Gibbon, Church, & Meck, 1984). Access to semantic memory may therefore be required to retrieve these representations for future use. If this were the case, efficiency of access to semantic memory may be important in forms of timing which require the retrieval of long-term memory representations of duration.

The dual-task studies reviewed above demonstrate the clear involvement of executive resources in timing. However, it should be noted, that most of the aforementioned studies used production or reproduction to assess timing. As a result, our understanding of the role of executive functions in timing is largely limited to the role of executive functions in these two motor timing tasks. This is problematic because different temporal tasks do not always demonstrate the same behavioural effects, perhaps indicating that different underlying mechanisms sub-serve each task (Baudouin, Vanneste, Isingrini and Pouthas, 2006, Baudouin, Vanneste, Pouthas and Isingrini, 2006, Block et al., 1998, Franssen and Vandierendonck, 2002, Gil and Droit-Volet, 2011, Mioni et al., 2013, Ogden and Jones, 2009, Ogden, Salominaite, Jones, Fisk and Montgomery, 2011, Ogden, Wearden, Gallagher and Montgomery, 2011, Ulbrich et al., 2007, Wearden et al., 2008, Wiener et al., 2009). For example, Baudouin, Vanneste, Isingrini, et al. (2006) observed that working memory capacity influences reproduction but not production performance. Given the differential role of working memory in different timing tasks, it is anticipated that temporal tasks will differentially recruit executive functions.

The current study sought to further explore how different executive functions (updating, switching, inhibition and access) may be recruited during interval timing. Of key interest was how resource recruitment may differ between different types of timing task. Unlike previous research using a dual-task interference paradigm (Brown, 2006, Brown et al., 2013, Ogden, Salominaite, Jones, Fisk and Montgomery, 2011, Rattat, 2010) an individual differences approach was taken. Individual differences in working memory and updating have been shown to influence children's and adults' temporal bisection performance (Zélanti and Droit-Volet, 2011, Zélanti and Droit-Volet, 2012) and adults' reproduction performance (Ulbrich et al., 2007). In a series of studies Zélanti and Droit-Volet, 2011, Zélanti and Droit-Volet, 2012 explored how children's working memory and attentional development influenced their timing ability. Zélanti and Droit-Volet (2012) observed that backward digit span and performance on a subset of the NEPSY (Korkman, Kirk, & Kemp, 1998) designed to test auditory attention and executive function (Auditory Attention and Response Set tasks) correlated with some elements of children's auditory bisection performance, but not with visual bisection. Performance on the same NEPSY subset also predicted temporal sensitivity for long durations (> 15 s) in both children and adults (Zélanti & Droit-Volet, 2011). Given the dual-task evidence for the recruitment of executive functions in timing, and the preliminary evidence that individual differences in executive and working memory performance affect timing, it is expected that individual differences in adults' executive capacity will influence timing performance.

Participants completed computer based tasks known to measure executive capacity for updating: computation span, switching: the number–letter task, inhibition: Random Letter Generation and access: The Chicago Word Fluency Test (CWFT). Critically, participants also completed three different timing tasks; temporal generalisation, reproduction and verbal estimation. Temporal generalisation, reproduction and verbal estimation were selected because of the varying demand they place on memory, attention and executive functions. Temporal generalisation requires participants to identify a previously learned standard duration from a range of comparison durations which are shorter, longer or equal in duration to the standard. Trial-by-trial retrieval of the standard from memory, whilst monitoring and maintaining the duration of comparisons in working memory, means that memory demands are high. Temporal generalisation may therefore recruit updating and access. Reproduction also places demands on memory; participants must maintain a representation of the duration to-be-reproduced whilst simultaneously monitoring the duration of their reproduction, thus updating and access may be recruited. Reproduction also presumably requires a degree of switching, in that participants must switch between monitoring the duration of their reproduction with their memory representation of the to-be-reproduced duration. Verbal estimation requires participants to monitor the duration of a stimulus and then provide a numeric estimation of the duration. Relative to temporal generalisation and reproduction, verbal estimation may be perceived as placing less demand on attention, memory and executive functions because participants are only processing and maintaining one stimulus at a time. Thus, verbal estimation may only recruit updating resources to monitor the to-be-timed stimulus. However, Wearden, Todd, and Jones (2006) suggest that participants may make reference to long-term memory representations of durations when performing verbal estimation, for example, their representation of 1 s. It is therefore possible that access resources may also be recruited during verbal estimation.

These suggestions were explored by a) examining correlations between performance on executive tasks and timing tasks b) comparing the timing ability of low and high scorers on the executive tasks, and c) examining the relationship between executive performance and computer models of human timing using a version of the Modified Church and Gibbon model (MCG) developed by Wearden (1992) from earlier work by Church and Gibbon (1982).