The influence of focused-attention meditation states on the cognitive control of sequence learning

Highlights

• Sequence learning is influenced by preceding focused-attention meditation (FAM).

• FAM immediately promotes stimulus-based planning and less sequence acquisition.

• Delayed influences of FAM enhance sequence acquisition and efficient responding.

• FAM influences sequential behaviour through top-down control processes.

Abstract

Cognitive control processes influence how motor sequence information is utilised and represented. Since cognitive control processes are shared amongst goal-oriented tasks, motor sequence learning and performance might be influenced by preceding cognitive tasks such as focused-attention meditation (FAM). Prior to a serial reaction time task (SRTT), participants completed either a single-session of FAM, a single-session of FAM followed by delay (FAM+) or no meditation (CONTROL). Relative to CONTROL, FAM benefitted performance in early, random-ordered blocks. However, across subsequent sequence learning blocks, FAM+ supported the highest levels of performance improvement resulting in superior performance at the end of the SRTT. Performance following FAM+ demonstrated greater reliance on embedded sequence structures than FAM. These findings illustrate that increased top-down control immediately after FAM biases the implementation of stimulus-based planning. Introduction of a delay following FAM relaxes top-down control allowing for implementation of response-based planning resulting in sequence learning benefits.

Introduction

Sequential behavior prominently features in most of the daily activities we undertake throughout our lives. Fundamental to the success of our lives then is the acquisition of sequential actions and this ability, in turn, relies on how consciousness and cognitive processes shape learning, adaptation and representation (Cleeremans and Jimenez, 2002, Delevoye-Turrell and Bobineau, 2012, Keele et al., 2003, Langer, 2000). Beyond sequential actions, other forms of goal-directed behavior, including cognitive-oriented tasks, also utilise and consequently influence cognitive control mechanisms. Because of shared utilisation of cognitive control processes, it would seem reasonable to consider that sequence learning could be influenced by a preceding cognitive-oriented task such as meditation, which has been shown to influence attentional and top-down versus bottom-up regulatory processes. The present experiment was designed to test this proposal by examining the effects of meditation on subsequent sequence learning. To provide relevant theoretical background for this experiment we provide overview of current models of sequence learning and representation and meditation, the role of attentional processes in both and preliminary research investigating the effects of meditation on motor learning.

Cognitive control processes underlying sequence learning are needed to regulate attention, working memory, and other executive functions (e.g. response selection, conflict resolution, and task representation) are utilised to increase performance over a period of practice (Daltrozzo and Conway, 2014, Keele et al., 2003, Slagter et al., 2011). Several models have been proposed to describe cognitive control processes and how they influence performance characteristics of sequenced actions (Abrahamse et al., 2010, Abrahamse et al., 2013, Clegg et al., 1998, Daltrozzo and Conway, 2014, Keele et al., 2003, Kelly et al., 2003, Schwarb and Schumacher, 2012, Slagter et al., 2011). A prominent feature across these models is the extent to which task features are attended and utilised during the acquisition of the sequential action. For example, control of sequence performance can either emphasise stimulus-based or response-based planning strategies (Tubau, Hommel, & Lopez-Moliner, 2007). Stimulus-based planning relies on the presentation of a signal that is associated with a specific response within the sequence as a ‘prepared-reflex’ (Hommel, 2000). Thus, the stimulus is used to signal the appropriate response in an automated way but does not afford further elaboration within the context of the sequence (Tubau et al., 2007). The notion of stimulus-based planning is very similar to the idea of unidimensional representation proposed by Keele et al. (2003) whereby a limited input source is used to predict a future response. Response-based planning, in contrast, utilises information from the action plan as a source of control, treating the response as a link within the sequence representations (Hommel, 1996, Tubau et al., 2007). Response-based planning thus emphasises associations between the representations of actions and outcomes into internalised plans that meet the goals of learning rather than just stimulus-driven responses (Elsner and Hommel, 2001, Holland, 2008, Hommel, 1996). Therefore, under response-based planning, motor responses become optimised and automated with efficient management of cognitive control processes in planning, decision-making, error-detection and memory formation (Alvarez and Emory, 2006, Chiappe et al., 2000, Gallant, 2016, Kane et al., 1994). Here, Keele et al. (2003) similarly refer to use of a multidimensional representation, whereby sequence elements are based on associations across a number of task features and into higher levels of goal orientation and cognition.

Abrahamse et al. (2010) and Abrahamse et al. (2013) also describe multiple strategies underlying sequence performance. Their model specifically attributes the use of performance strategies to the extent to which the sequence representation has been developed over a period practice. During early stages of sequence learning, performance follows a reaction mode, which is similar to the notion of stimulus-based (Tubau et al., 2007) or unidimensional (Keele et al., 2003) strategies or what Clegg et al. (1998) refer to as low level encoding. As practice continues, performance begins to utilise an associative mode whereby sequence automatisation increases owing to an abstract-rule formation (Abrahamse et al., 2013, Franco and Destrebecqz, 2012, Grafton et al., 1998) or referred to as intermediate level associations (Clegg et al., 1998). Further practice results in increased sequence automatisation that results from sequence chunking whereby sequence elements have been abstracted into long-term memory as sub-units within the overall structure of a sequence (Abrahamse et al., 2010, Abrahamse et al., 2013, Clegg et al., 1998, Jimenez, 2008, Koch and Hoffmann, 2000, Verwey and Abrahamse, 2012, Verwey and Wright, 2014). The availability of the chunking mode allows sequence production to no longer rely on stimulus-response mapping given sequence sub-units are preloaded in working memory, which allows the performer to anticipate the upcoming responses within the sequence (Jimenez, 2003, Jimenez, 2008). The formation of sequence chunks might conceivably support response-based planning (Tubau et al., 2007) based on high level of multidimensional associations (Keele et al., 2003) and high level abstract representation (Clegg et al., 1998) to perform speeded and accurate sequential action. Cognitive control is responsible for directing attention towards task-relevant information and inhibiting competing or irrelevant information (Gallant, 2016, Lyon and Krasnegor, 1996, Miyake et al., 2000) and as such, cognitive control would influence whether sequence learning and representation is based on stimulus-based or response-based planning.

Meditation typically involves a set of guided instructions that establishes a state of cognitive control influencing attention, performance monitoring and working memory (Cahn and Polich, 2006, Debarnot et al., 2014, Loizzo, 2014, Moore and Malinowski, 2009, Slagter et al., 2007, Zeidan et al., 2010). Accordingly, a period of meditation training has been reported to improve performance on tasks that assess attention, interference control and working memory function (Gallant, 2016, Hasenkamp et al., 2012, Malinowski, 2013, Tang et al., 2015, Tang et al., 2007). To explain these effects, current theoretical models emphasise attention as the central component influenced by meditation (Moore et al., 2012, Moore and Malinowski, 2009). It is through its influence on attention control processes that meditation is thought to lend benefits for the function of subsequent processing stages such as working memory (Malinowski, 2013, Tang and Posner, 2009).

As meditation is thought to have a primary influence on attention, meditation styles can be described based on the manner by which the technique influences the control of attention (Lippelt et al., 2014, Lutz et al., 2008). Focused attention meditation (FAM) is a meditation style that is characterised by the goal of maintaining sustained attention on a specific object (e.g., breathing) (Lippelt et al., 2014, Lutz et al., 2008, Slagter et al., 2011). Thus, FAM utilises top-down cognitive control processes that constrain attention in a narrow or convergent manner in order to sustain attention on a target object (Colzato et al., 2016, Lutz et al., 2015, van Leeuwen et al., 2012). Although FAM places an emphasis on sustaining attention on a target object, to meet this goal, other cognitive control processes are utilised as it has been reported that even experienced meditators become distracted about once every 80 s (Hasenkamp & Barsalou, 2012). Thus, sustained attention has been proposed to be but one part of a top-down oriented cognitive control cycle that is implemented in FAM (Gallant, 2016, Hasenkamp and Barsalou, 2012, Hasenkamp et al., 2012, Malinowski, 2013, Miyake et al., 2000, Posner and Petersen, 1990, Posner and Rothbart, 1998). Specifically, in addition to maintaining sustained attention on goal-relevant stimuli, top-down control also utilises processes associated with vigilance and monitoring of attention allocation, detection of distraction from irrelevant stimuli, inhibiting or disengaging from irrelevant stimuli and reengaging or refocusing attention on relevant stimuli (Gallant, 2016, Hasenkamp and Barsalou, 2012, Malinowski, 2013, Miyake et al., 2000, Posner and Rothbart, 1998).

Recent work has demonstrated that a single session of FAM establishes top-down cognitive control states that instantaneously bias attention allocation policies on subsequent tasks (Baas et al., 2014, Colzato et al., 2012, Colzato et al., 2015, Colzato et al., 2015, Lippelt et al., 2014). For example, a single session of FAM with meditation naïve participants has been reported to result in greater suppression of task-irrelevant visual stimuli (Colzato et al., 2016) and higher levels of performance adaptation following exposure to stimuli which where incongruent with the goal response (Colzato, Sellaro, Samara, & Hommel, 2015). In addition, a single session of FAM has been reported to result in larger attentional blink responses when compared to the effects of a meditation style called open monitoring meditation that promotes more diffuse allocation of attention (Colzato, Sellaro, Samara, Baas, et al., 2015). The larger attentional blink following a single session of FAM is consistent with the notion that this meditation style can instantaneously promote cognitive control states that render attention more selective to target stimuli. These recent reports highlight the enduring effect that a single session of meditation can have on subsequent unrelated goal-oriented tasks (Lippelt et al., 2014, Malinowski, 2013, Tang et al., 2015). However, the duration by which a single session of meditation can provide carry-over effects on subsequent tasks is not known. Moreover, the extent by which meditation states provide carry-over effects on various task types, including those that emphasise cognitive processes associated with sequence learning, is not well understood.

That FAM can establish top-down cognitive control states that influence the performance of subsequent tasks, suggests that the utilisation of stimulus-based or response-based strategies in sequence learning (Tubau et al., 2007) can be biased by a preceding single session of FAM. For example, instantaneous effects of FAM might mean that top-down control states narrow attention to the extent that the stimulus is predominantly utilised in signalling the appropriate response (Hommel, 2000, Keele et al., 2003, Tubau et al., 2007). van Leeuwen et al. (2012) have demonstrated that FAM promotes faster responding to local task-relevant stimuli, at least in those experienced within the meditation. Control strategies that emphasise attention allocation to task-relevant stimuli might benefit those in the early stages of sequence learning where performance is thought to predominantly involve low level stimulus encoding (Abrahamse et al., 2010, Abrahamse et al., 2013, Clegg et al., 1998, Hommel, 2000, Keele et al., 2003, Tubau et al., 2007).

In addition to influencing control strategies that allocate attention to task-relevant stimuli, FAM may also influence regulation of working memory, since the function of the latter would be required in order to undertake processes such as monitoring attention allocation and reengagement of task-relevant stimuli involved in the cyclic process of focusing attention (Gallant, 2016, Hasenkamp and Barsalou, 2012, Malinowski, 2013, Miyake et al., 2000, Posner and Petersen, 1990). Working memory encodes, processes, and maintains information governed by cognitive control processes (Baddeley et al., 1999, Baddeley and Hitch, 1974, Burnham et al., 2014, Friedman and Miyake, 2004, Miyake and Shah, 1999). Moreover, attention and working memory are interdependent during performance of goal-directed tasks (Koechlin & Summerfield, 2007), particularly in the presence of task-irrelevant information (Hutchinson and Turk-Browne, 2012, James, 1890, Lavie, 2006, Lavie et al., 2004, Mizrak and Oztekin, 2016). The presence of task-irrelevant information contributes to the overall attentional load and so attention relies on working memory to ascertain task-relevant information from a background of irrelevant or distracting information (De Kleine and Van der Lubbe, 2011, Lavie and Torralbo, 2010). Thus, it is conceivable that the instantaneous effects of FAM could result in the utilisation of response-driven strategies during sequence learning as opposed to stimulus-driven strategies, because FAM utilises working memory in a way that primes inclusion of action related information during acquisition of the sequence. Support for the idea that FAM might orient response-driven sequence learning is based on a demonstration of experienced meditators having greater capacity in using action oriented information for accurate response selection than non-meditators (Delevoye-Turrell & Bobineau, 2012). Whether long term influences of FAM on response-driven strategies are also reflected in instantaneous effects has not yet been tested.

The influence of FAM on working memory function is particularly relevant to implementing response-based forms of planning during sequence learning since this kind of planning requires capacity to link elements of the response sequence (Hommel, 1996, Tubau et al., 2007). Forming response associations or motor chunks, early in sequence learning would be advantageous since the learner would utilise internalised plans to perform sequenced action rather than relying on sensory input to drive each component of the sequence (Abrahamse et al., 2013, Clegg et al., 1998, Elsner and Hommel, 2001, Franco and Destrebecqz, 2012, Grafton et al., 1998, Holland, 2008, Hommel, 1996). Moreover, response-based planning allows working memory to better support attention in handling the informational load of the task since the internalised sequence structure representation can be used to select an upcoming response as opposed to having to attend to an external stimulus (Jimenez et al., 2011, Mizrak and Oztekin, 2016).

Whether FAM promotes stimulus-based or response-based planning forms of sequence learning might be influenced by the temporal proximity by which FAM precedes sequential learning. The temporal dynamics of cognitive control (Koechlin et al., 2003, Koechlin and Summerfield, 2007) might determine if sequence learning is biased such that attention is narrowly placed only on proximal information (i.e., stimulus-based planning) or is biased to rely more on working memory thus allowing for association between immediate and past action information in predicting upcoming responses (i.e., response-based planning). Thus, if sequence learning immediately follows FAM, the learner might be biased to only utilise the immediate stimulus in selecting each element of the sequence. If there is a delay between FAM and sequence learning, top-down constraining of attention might be weakened to the extent that a wider array of information residing in working memory is processed even if this information does not appear to directly relate to immediate task performance. The weakening of top-down cognitive control has been shown to provide benefits for other tasks that do not require a narrow attention such as learning of regularities in sequence learning and development of memory in tasks (Amer et al., 2016, Borragan et al., 2016). This would mean that placing an interval of time between completion of FAM and initiation of sequence learning might promote greater response-based planning.

Unfortunately, what little previous work exists that addresses the instantaneous effects of meditation on motor learning does not provide for a clear answer with respect to whether FAM promotes stimulus or response-driven strategies. Appelle and Oswald (1974) reported shorter auditory simple reaction time following FAM, which is consistent with the idea that FAM biases utilisation of stimulus-based planning. However, FAM has also been shown to either diminish performance on subsequent stimulus-oriented tasks such as visual tracking (Williams & Herbert, 1976) or to have no effect on mirror tracing performance (Williams, 1978). Support for the potential of FAM to promote response-based planning relates to reports of reduced tapping task errors (Telles, Hanumanthaiah, Nagarathna, & Nagendra, 1993), increased motor response speed (Dash & Telles, 1999) or improve grip force accuracy (Delevoye-Turrell & Bobineau, 2012). The answer to the instantaneous effects of FAM on sequence learning is obscured by the fact that the previous work has not specifically investigated sequence learning and these studies have included experienced meditators, which raises uncertainty about immediate versus long-term effects of FAM.

The goal of the current experiment was to test the effects of a single session of FAM on subsequent motor sequence learning. We employed the serial reaction time task (SRTT; Nissen & Bullemer, 1987) since this paradigm allows for determination if sequence learning relies primarily on stimulus-based planning under the notion of general practice effects (Abrahamse & Noordzij, 2011) or response-based planning under the notion of sequence-specific learning (Robertson, 2007, Willingham, 1999, Willingham et al., 2000). We predicted that if FAM biases top-down cognitive control to narrow attention such that only stimuli are selectively attended to then improvements in SRTT performance should align with a pattern of general practice effects where decreasing response latencies are not attributable to a sequence structure. In contrast, if FAM promotes utilisation of working memory as part of the top-down cognitive control strategy then improvements in SRTT performance should align with a pattern of sequence-specific learning where improvements in response latencies are only evident when the sequence structure established during practice is available. Furthermore, we predicted that if FAM promotes response-based planning to the extent that greater utilisation of working memory promotes formation of motor memory for the practiced sequence, then performance on the practiced sequence should be more resilient to the introduction of an interfering sequence structure. In addition, if FAM promotes motor memory formation for the sequence through utilisation of response-based strategies, then individuals would be expected to be better able to reproduce sequence structure chunks when asked to reproduce the sequence at the completion of practice.