Can frequent long stimulus onset ansynchronies (SOAs) foster the representation of two separated task-sets in dual-tasking?

Recent findings suggest that in dual-tasking the elements of the two tasks are associated across tasks and are stored in a conjoint memory episode, meaning that the tasks are not represented as isolated task-sets. In the current study, we tested whether frequent long stimulus onset ansynchronies (SOAs) can foster the representation of two separated task-sets thereby reducing or even hindering participants to generate conjoint memory episodes-compared to an integrated task-set representation induced by frequent short SOAs. Alternatively, it is conceivable that conjoint memory episodes are an inevitable consequence of presenting two tasks within a single trial. In two dual-task experiments, we tested between consecutive trials whether repeating the stimulus-response bindings of both tasks would lead to faster responses than repeating only one of the two tasks' stimulus-response bindings. The dual-task consisted of a visual-manual search task (VST) and an auditory-manual discrimination task (ADT). Overall, the results suggest that, after processing two tasks within a single trial, generating a conjoint memory episode seems to be a default process, regardless of SOA frequency. However, the respective SOA frequency affected the participants' strategy to group the processing of the two tasks or not, thereby modulating the impact of the reactivated memory episode on task performance.

Across-task binding: The development of a representation in learning a continuous movement sequence.

Across-task binding is defined as the stimulus/response of one task being linked to the response of another task. The purpose of the present experiment was to determine across-task binding in a continuous movement sequence task with an auditory task of high and low pitch tones and the development of a movement sequence representation. According to the two systems theory of sequence learning, we expected that the developed representation in the across-task binding context relies on the multi-dimensional system rather than on the unidimensional system which is restricted to a set of modules where each module processed information along one task/dimension. An inter-manual transfer design was used to disentangle the sequence representations. The mirror transfer test required the same pattern of muscle activation and joint angles (motor coordinates) in the contralateral limb as experienced during the acquisition phase, while in the non-mirror transfer test, the visual-spatial locations (spatial coordinates) of the target waveform were reinstated. The main finding was that consistently combining visual-spatial positions in a sequence and auditory dimensions such as the tone pitch does not rely on a multidimensional system as predicted by the two-systems theory.

Concurrent visual sequence learning.

Many researchers in the field of implicit statistical learning agree that there does not exist one general implicit learning mechanism, but rather, that implicit learning takes place in highly specialized encapsulated modules. However, the exact representational content of these modules is still under debate. While there is ample evidence for a distinction between modalities (e.g., visual, auditory perception), the representational content of the modules might even be distinguished by features within the same modalities (e.g., location, color, and shape within the visual modality). In implicit sequence learning, there is evidence for the latter hypothesis, as a stimulus-color sequence can be learned concurrently with a stimulus-location sequence. Our aim was to test whether this also holds true for non-spatial features within the visual modality. This has been shown in artificial grammar learning, but not yet in implicit sequence learning. Hence, in Experiment 1, we replicated an artificial grammar learning experiment of Conway and Christiansen (2006) in which participants were supposed to learn color and shape grammars concurrently. In Experiment 2, we investigated concurrent learning of sequences with an implicit sequence learning paradigm: the serial reaction time task. Here, we found evidence for concurrent learning of two sequences, a color and shape sequence. Overall, the findings converge to the assumption that implicit learning might be based on features.

Parallel Acquisition of Uncorrelated Sequences does Not Provide Firm Evidence for a Modular Sequence-Learning System.

Dual-systems theories of sequence learning assume that sequence learning may proceed within a unidimensional learning system that is immune to cross-dimensional interference because information is processed and represented in dimension-specific, encapsulated modules. Important evidence for such modularity comes from studies investigating the absence or presence of interference between multiple uncorrelated sequences (e.g., a sequence of color stimuli and a sequence of motor keypresses). Here we question the premise that the parallel acquisition of uncorrelated sequences provides convincing evidence for a modularized learning system. In contrast, we demonstrate that parallel acquisition of multiple uncorrelated sequences is well predicted by a computational model that assumes a single learning system with joint representations of stimulus and response features.

Action-effect knowledge transfers to similar effect stimuli.

The ability to anticipate the sensory consequences of our actions (i.e., action-effects) is known to be important for intentional action initiation and control. Learned action-effects can select the responses that previously have been associated with them. What has been largely unexplored is how learned action-effect associations can aid action selection for effects that have not previously associated with an action but are similar to learned effects. In two studies, we aimed to show that when presented new, unknown action-effects, participants select the responses that have previously been associated with similar action-effects. In the first study (n = 27), action-effect similarity was operationalized via stimuli belonging to the same or different categories as the previously learned action-effects. In the second study (n = 31), action-effect similarity was realized via stimuli that require comparable motor responses in real life. Participants first learned that specific responses are followed by specific visual effect stimuli. In the test phase, learned effect stimuli, new but similar effect stimuli and new but dissimilar effect stimuli were presented ahead of the response. The findings revealed that both learned effect stimuli and new similar effect stimuli affected response times, whereas new dissimilar effects did not. When a learned or a new similar effect was followed by a learned response, compared to an unlearned response, the responses were faster. We interpret these findings in terms of action-effect learning. The action-effect once bound to an action is used to select an action if a similar effect for which no action has been learned yet is presented. However, it is noteworthy that, due to our design, other explanations for the found transfer are conceivable. We address these limitations in the General Discussion.

Element-level features in conjoint episodes in dual-tasking.

The usual way of thinking about dual-tasking is that the participants represent the two tasks separately. However, several findings suggest that the participants rather seem to integrate the elements of both tasks into a conjoint episode. In three experiments, we aimed at further testing this task integration account in dual-tasking. To this end, we investigated how the processing of the previous Trial n-1 shapes the processing of the current Trial n. We observed performance benefits when the stimulus-response mappings of both tasks repeat in consecutive trials (full repetition: FR) as compared to when only one such mapping repeats (partial repetition: PR). In particular, our experiments focused on the question which elements of the two tasks in dual-tasking might be bound together. For this purpose, in Experiments 1 and 2, all participants performed a dual-task consisting of a visual-manual search task (VST) and an auditory-manual discrimination task (ADT). In the VST the stimulus-response mappings were variable, so that none of the stimuli of this task systematically predicted a certain response. In Experiment 1, the stimuli and responses of the VST were either both repeated or both changed in consecutive trials. In Experiment 2, we removed the stimulus repetitions in the VST and only the responses repeated across trials. In Experiment 3, we changed the ADT into a visual-auditory matching task (VAMT) with variable stimulus-response mappings, so that in both tasks only the responses repeated across trials. In Experiments 1 and 2, we observed better performance for FR than for PR, while this difference disappeared in Experiment 3. Together, the results suggest that the stimulus of one task is sufficient to retrieve the entire episode from the previous trial.

High fluency can improve recognition sensitivity based on learned metacognitive expectations.

Fluency of processing has shown to influence recognition judgments. Fluency most commonly induces a liberal response bias to judge fluently processed information as well-known because knowledge of a high correlation between the frequency of encounters, memory strength, and thus fluency of processing has been acquired in the past. In this study, we aimed to show that high fluency can increase recognition judgment sensitivity as well if the participants had encountered fluent and non-fluent processing during training. Thirty-three participants have been trained with a 12-element sequence in a serial reaction time task. During training, the response stimulus interval alternated block-wise between constant (fluent) and variable (non-fluent). Participants showed a higher capability of discriminating between old and new test sequences under fluent than under non-fluent test conditions. Furthermore, participants did not show any liberal or conservative bias after they have been trained with alternating fluency.

Given the option, people avoid incongruent responses in a dual-tasking situation.

While past work on how people can optimize dual-tasking has focused on strategic timing (i.e., when to select responses), little is known about the extent to which people can optimize dual-tasking by taking care of which responses they select. Here we test whether spatial (in)congruency influences response selection in free-choice trials. In two experiments, we combined two visual-manual tasks with spatial stimulus- and response characteristics: Participants responded to the stimulus words "left" and "right" in a forced choice task and responded "up", "down", "left" or "right" with an arrow-key to either a free choice prompt or an X located at the respective position. In Experiment 1, participants reduced the proportion of incongruent pairs of responses (i.e., left in one and right in the other task). In Experiment 2, we found that such flexibility in response selection also holds in more constrained environments: Within runs of four trials the free-choice options were continuously reduced based on the responses already given. The combined results of Experiments 1 and 2 suggest that response selection in free choice trials is driven by performance optimization beyond response conflict.

The interplay between unexpected events and behavior in the development of explicit knowledge in implicit sequence learning.

Some studies in implicit learning investigate the mechanisms by which implicitly acquired knowledge (e.g., learning a sequence of responses) becomes consciously aware. It has been suggested that unexpected changes in the own behavior can trigger search processes, of which the outcome then becomes aware. A consistent empirical finding is that participants who develop explicit knowledge show a sudden decrease in reaction times, when responding to sequential events. This so called RT-drop might indicate the point of time when explicit knowledge occurs. We investigated whether an RT-drop is a precursor for the development of explicit knowledge or the consequence of explicit knowledge. To answer this question, we manipulated in a serial reaction time task the timing of long and short stimulus-onset asynchronies (SOA). For some participants, the different SOAs were presented in blocks of either long or short SOAs, while for others, the SOAs changed randomly. We expected the participants who were given a blocked presentation to express an RT-drop because of the predictable timing. In contrast, randomly changing SOAs should hamper the expression of an RT-drop. We found that more participants in the blocked-SOA condition than in the random-SOA condition showed an RT-drop. Furthermore, the amount of explicit knowledge did not differ between the two conditions. The findings suggest that the RT-drop does not seem to be a presupposition to develop explicit knowledge. Rather, it seems that the RT-drop indicates a behavioral strategy shift as a consequence of explicit knowledge.

Task-separation in dual-tasking: How action effects support the separation of the task streams.

Dual-task costs might reflect a direct consequence of confusions on the task-set level as both tasks are integrated into a single task-set, instead of two separate task-sets. In order to prevent this integration-driven interference, the two task streams have to be separated. Under three experimental conditions we investigated whether in a dual-task setting such separated task processing can be elicited by providing separated action effects for both tasks. Building on the finding that implicit sequence learning is hampered under dual-tasking conditions, we used it as an indicator for successfully separated task processing. To this end, we compared the implicit sequence learning effect under a single-task condition (baseline condition) to that under a dual-task condition either with separated action effects (action-effect condition) or with conjoined error feedback at the end of each trial (feedback condition): Learning should be unaffected under the former and reduced under the latter experimental condition. In all three conditions, the participants performed a visual-manual and an auditory-manual discrimination task which were always presented concurrently. The results show that, compared to the single-task condition, under dual-task conditions implicit sequence learning is hampered when only a conjoined error feedback is given, but is largely unaffected by dual-tasking when separated action effects are presented. Overall this suggests that whenever two tasks in a dual-task situation appear concurrently, they seem to get integrated into a single task-set by default. Yet, manipulations like the presentation of separated action effects could help to elicit a separated task processing, thus strengthening the representation as two separate task-sets.

Learning of across- and within-task contingencies modulates partial-repetition costs in dual-tasking.

Dual-task costs might result from confusions on the task-set level as both tasks are not represented as distinct task-sets, but rather being integrated into a single task-set. This suggests that events in the two tasks are stored and retrieved together as an integrated memory episode. In a series of three experiments, we tested for such integrated task processing and whether it can be modulated by regularities between the stimuli of the two tasks (across-task contingencies) or by sequential regularities within one of the tasks (within-task contingencies). Building on the experimental approach of feature binding in action control, we tested whether the participants in a dual-tasking experiment will show partial-repetition costs: they should be slower when only the stimulus in one of the two tasks is repeated from Trial n - 1 to Trial n than when the stimuli in both tasks repeat. In all three experiments, the participants processed a visual-manual and an auditory-vocal tone-discrimination task which were always presented concurrently. In Experiment 1, we show that retrieval of Trial n - 1 episodes is stable across practice if the stimulus material is drawn randomly. Across-task contingencies (Experiment 2) and sequential regularities within a task (Experiment 3) can compete with n - 1-based retrieval leading to a reduction of partial-repetition costs with practice. Overall the results suggest that participants do not separate the processing of the two tasks, yet, within-task contingencies might reduce integrated task processing.

What triggers explicit awareness in implicit sequence learning? Implications from theories of consciousness.

This article aims to continue the debate on how explicit, conscious knowledge can arise in an implicit learning situation. We review hitherto existing theoretical views and evaluate their compatibility with two current, successful scientific concepts of consciousness: The Global Workspace Theory and Higher-Order Thought Theories. In this context, we introduce the Unexpected Event Hypothesis (Frensch et al., Attention and implicit learning, John Benjamins Publishing Company, 2003) in an elaborated form and discuss its advantage in explaining the emergence of conscious knowledge in an implicit learning situation.

No reduction of between-task interference in a dual-task with a repeating sequence of SOAs.

A frequent observation in dual-tasking is that spatially or conceptually (in)compatible Task 2 response features can interfere with responses in Task 1 (backward crosstalk effect; BCE). Such between-task interference is, at least to some degree, under strategic control. It has been shown that the size of the BCE can be modulated by instructions, contextual regularities, recent experience of conflict, and motivational factors. Especially large temporal task overlap (i.e., short stimulus onset asynchrony, SOA) represents a condition of potentially high levels of between-task interference. Accordingly, Fischer and Dreisbach (2015) showed that specific stimuli, associated with mostly short SOAs, were able to reduce the size of the BCE. In the present study, we investigated whether a regular sequence of SOAs can also be used for contextual regulation of the BCE. In a dual-task with spatially (in)compatible hand- and foot-responses, we implemented a repeating sequence of three SOAs. If participants learned this sequence and used it for task shielding, the BCE should decrease over time in the sequence blocks, but should increase in a subsequent random block. However, this prediction was not supported in two experiments (N = 32 each). Instead, the size of the BCE was constant across all blocks (BFs10 < 1 for the respective interactions). This is an important result, as it points at the necessity to discover the appropriate conditions allowing implicit SOA sequence learning and to further investigate whether or how the resulting implicit sequence knowledge can serve shielding against between-task interference.

Why Does Dual-Tasking Hamper Implicit Sequence Learning?

Research on the limitations of dual-tasking might profit from using setups with a predictable sequence of stimuli and responses and assessing the acquisition of this sequence. Detrimental effects of dual-tasking on implicit sequence learning in the serial reaction time task (SRTT; Nissen & Bullemer, 1987) - when paired with an uncorrelated task - have been attributed to participants' lack of separating the streams of events in either task. Assuming that co-occurring events are automatically integrated, we reasoned that participants could need to first learn which events co-occur, before they can acquire sequence knowledge. In the training phase, we paired an 8-element visual-manual SRTT with an auditory-vocal task. Afterwards, we tested under single-tasking conditions whether SRTT sequence knowledge had been acquired. By applying different variants of probabilistic SRTT-tone pairings across three experiments, we tested what type of predictive relationship was needed to preserve sequence learning. In Experiment 1, where half of the SRTT-elements were paired to 100% with one specific tone and the other half randomly, only the fixedly paired elements were learned. Yet, no sequence learning was found when each of the eight SRTT-elements was paired with tone identity in a 75%-25% ratio (Experiment 2). Sequence learning was, however, intact when the 75%-25% ratio was applied to the four SRTT target locations instead (Experiment 3). The results suggest that participants (when lacking a separation of the task representations while dual-tasking) can learn a sequence inherent in one of two tasks to the extent that across-task contingencies can be learned first.

Parallel and serial task processing in the PRP paradigm: a drift-diffusion model approach.

Even after a long time of research on dual-tasking, the question whether the two tasks are always processed serially (response selection bottleneck models, RSB) or also in parallel (capacity-sharing models) is still going on. The first models postulate that the central processing stages of two tasks cannot overlap, producing a central processing bottleneck in Task 2. The second class of models posits that cognitive resources are shared between the central processing stages of two tasks, allowing for parallel processing. In a series of three experiments, we aimed at inducing parallel vs. serial processing by manipulating the relative frequency of short vs. long SOAs (Experiments 1 and 2) and including no-go trials in Task 2 (Experiment 3). Beyond the conventional response time (RT) analyses, we employed drift-diffusion model analyses to differentiate between parallel and serial processing. Even though our findings were rather consistent across the three experiments, they neither support unambiguously the assumptions derived from the RSB model nor those derived from capacity-sharing models. SOA frequency might lead to an adaptation to frequent time patterns. Overall, our diffusion model results and mean RTs seem to be better explained by participant's time expectancies.

Origami folding: Taxing resources necessary for the acquisition of sequential skills.

Sequential skill learning with practice is fundamental to human activity (e.g., tying shoes). Given the lack of prior knowledge in most participants, Origami folding is a promising task to study the acquisition of a sequential skill. While previous Origami folding studies mainly dealt with the question, which forms of instruction can lead to better learning outcomes, we employ a dual-task approach to test which resources are necessary for folding and for improvement with practice. Participants (N = 53) folded five Origami figures for four times each, which were randomly paired with five types of secondary tasks to cause either cognitive (verbal vs. visuospatial) or motoric (isochronous vs. non-isochronous tapping) memory load or none (control condition). Origami performance showed a typical learning curve from Trial 1 (first run of folding the figure) to Trial 4 (fourth run of folding the same figure). We tested for a dissociation between variants of memory load influencing Origami folding performance vs. the variants influencing learning (i.e. change in performance across practice). In line with theories suggesting that learning operates on the level that (at a given point in practice) demands the most control, we did not observe cases where a dual-task variant influenced performance while it did not affect learning. Memory load from the cognitive visuospatial secondary task as well as the isochronous tapping secondary task interfered with improvement in Origami folding with practice. This might be due to the use of visuospatial sketchpad and absolute timing mechanism during the acquisition of Origami folding.

Sequential modulation of across-task congruency in the serial reaction time task.

While detrimental effects of dual-tasking on the acquisition and usage of sequence knowledge in the Serial Reaction Time Task (SRTT) have been attributed to the integration of regularly and randomly sequenced events, direct evidence for across-task integration has been scarce. In two experiments, we paired two spatial tasks: A visual-manual SRTT (serial reaction time task) of length 4 and a two-choice task with random order of stimuli. We reasoned that across-task integration might result in SRTT- and two-choice task events being stored and retrieved together. Therefore, RT might be influenced by current congruency as well as by whether it is repeated or altered from the congruency level encountered for this SRTT event in the last loop of the sequence. We observed such a modulation in two experiments, suggesting that across-task integration indeed takes place and that the detrimental effect of dual-tasking on sequence learning is, at least in parts, based on across-task integration.

Feature codes in implicit sequence learning: perceived stimulus locations transfer to motor response locations.

An important question in implicit sequence learning research is how the learned information is represented. In earlier models, the representations underlying implicit learning were viewed as being either purely motor or perceptual. These different conceptions were later integrated by multidimensional models such as the Dual System Model of Keele et al. (Psychol Rev 110(2):316-339, 2003). According to this model, different types of sequential information can be learned in parallel, as long as each sequence comprised only one single dimension (e.g., shapes, colors, or response locations). The term dimension, though, is underspecified as it remains an open question whether the involved learning modules are restricted to motor or to perceptual information. This study aims to show that the modules of the implicit learning system are not specific to motor or perceptual processing. Rather, each module processes an abstract feature code which represents both response- and perception-related information. In two experiments, we showed that perceiving a stimulus-location sequence transferred to a motor response-location sequence. This result shows that the mere perception of a sequential feature automatically leads to an activation of the respective motor feature, supporting the notion of abstract feature codes being the basic modules of the implicit learning system. This result could only be obtained, though, when the task instructions emphasized the encoding of the stimulus-locations as opposed to an encoding of the color features. This limitation will be discussed taking into account the importance of the instructed task set.

Sequence Knowledge on When and What Supports Dual-Tasking.

The constraints in overlapping response selection have been established in dual-tasking studies with random sequence of stimuli and responses as well as random stimulus onset asynchrony (SOA). While this approach makes it possible to control for advance activation of upcoming stimuli or responses, it leaves open whether such preparatory processing can indeed influence dual-task performance. We investigated whether and how the sequence of stimuli and responses and the sequence of SOAs can be learned and used under dual-tasking. In each trial, participants (N = 28 in Experiment 1 and N = 30 in Experiment 2) were first presented with a random two-choice task followed by a four-choice Serial Reaction Time Task (SRTT), presented in a sequence of length four (position sequence). The SOA (timing) sequence also had length four. In test phases, one or both of the sequences were randomized. Results showed that both position and timing sequences were learned and supported dual-task performance, suggesting that predictive processing with respect to timing and identity of stimuli and responses can help to circumvent the response selection bottleneck constraints. Furthermore, in contrast to previous work on acquisition of interval sequences in single tasking, we found that the sequence of what (i.e. stimulus) and the sequence of when (i.e. interval between two tasks) contributed independently to performance.

More Than Hitting the Correct Key Quickly.

Many studies have documented that multitasking reduces Response Time (RT) indicators of implicit sequence learning as well as the expression of acquired sequence knowledge in RT benefits. In these tasks it is only relevant that the correct key is hit quickly, not where it is hit. We explored how variability in response location is influenced by (a) breaking a repeating sequence of target locations, (b) multitasking demands in the current trial, and (c) presence of multitasking in the block. Participants performed a Serial Reaction Time Task (SRTT) on a touchscreen while shutting down a beep tone by pressing the space bar with their non-dominant hand (throughout Experiment 1 and in the second half of Experiment 2). The first-order sequence of four response locations on the screen was broken by off-sequence deviants in 1/6th of the trials. Our results show a dissociation between RT and response location variability. While the effect of breaking the sequence on RT was larger under single- than under multitasking, breaking the sequence only led to an increase in response location variability under multitasking. Experiment 3 suggested that the impact of sequence knowledge on either aspect of performance in the SRTT is limited by interference from an additional task.