Functional characteristics of instructed and practiced task-sets.
Friday, September 30th, 2011 [08:30 - 10:30]
SY_06. Functional characteristics of instructed and practiced task-sets
Liefooghe, B. 1 , Wenke, D. 2 & Dreisbach, G. 3
1 Ghent University
2 Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig
3 University of Regensburg
Goal-directed behavior is often assumed to be based on a task-set, which is a representation of the control settings for processes such as stimulus identification, response selection and response execution. Although task-sets play an important role in many theories of goal-directed behavior, their functional characteristics remain rather vague. The present symposium will focus on research with a direct interest on the functional characteristics of task-sets that are newly formed on the basis of task instructions, on the one hand, and of task-sets of tasks that have already been practiced on the other hand.
TALKS
SY_06.1 - Functional Characteristics of the instruction-based task-rule congruency effect
Liefooghe, B. 1 , Wenke, D. 2, 1 & De Houwer, J. 1
1 Ghent University
2 Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig
The present study investigated the functional characteristics of task-sets that were never applied before and that were only formed on the basis of instructions. We tested if such task-sets could elicit a task-rule congruency effect, which consists of the automatic activation of responses in the context of another task. A series of experiments indicate that task-sets, which are formed on the basis of instructions, can elicit a task-rule congruency effect. A finding challenging recent accounts of this effect. Additional experiments further investigated the functional properties of this instruction-based task-rule congruency effect. Implications for research on task-set formation and for research on the implementation of instructions are considered.
SY_06.2 - Object selection in declarative working memory is analogous to response selection in procedural working memory
Oberauer, K. , Souza da Silva, A. , Druey, M. & Gade, M.
University of Zurich
We propose that working memory (WM) is a system for making available representations that are relevant for cognition and action. Declarative WM makes available the objects of (cognitive) actions, whereas procedural WM holds representations of the planned actions themselves, that is, task sets. We assume that declarative and procedural WM operate by analogous mechanisms. We present four experiments testing this hypothesis, focusing on selection of items in declarative WM and of responses in procedural WM. In sequences of speeded choice tasks, selecting the same response on successive trials speeds up responding. This response repetition benefit is turned into a cost when people switch the task from one trial to the next. We show analogous pattern for the selection of items in declarative WM. Repeated access to the same digit in working memory resulted in a benefit, which turned into a cost when people switched between two lists in successive trials, and the same digit had to be accessed in different positions in the two sets. The final experiment shows the same pattern of repetition costs and benefits in a task-switch paradigm designed in exact analogy to the declarative WM task. The results are explained by the inhibition of items and responses when de-selected, together with temporary strengthening of bindings of items and responses to their retrieval cues.
SY_06.3 - Task sets are modality specific
Stephan, D. & Koch, I.
RWTH Aachen University
Current theories of cognitive control assume that a task set is needed to perform a certain task. Thus, it reflects the mental representation of the task and involves the necessary cognitive processes, the representation of the set of the relevant stimuli, the required responses and the mapping between these stimuli and responses. In general, task sets are described to be abstract and amodal. We report experiments in which we systematically examine the influence of modality specific factors on task sets using task switching methodology. More specifically, we examined the role of compatibility of input and output (I-O) modality mappings in task switching. Our experiments included switching between auditory-vocal and visual-manual tasks (compatible) and between auditory-manual and visual-vocal tasks (incompatible). Overall, the resulting switch costs were smaller when participants switched between compatible tasks compared to when they switched between incompatible tasks. Reduced switch costs in compatible tasks may be due to special linkages between input and output modalities, whereas incompatible tasks increase cross-talk, presumably due to dissipating interference of correct and incorrect response modalities. These special linkages are part of the task set. Our data suggest that task sets are modality specific rather than being entirely abstract and amodal!
SY_06.4 - Task Shielding and its relaxation during task switching
Dreisbach, G. 1 & Wenke, D. 2, 3
1 University of Regensburg
2 Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig
3 Ghent University
Task sets in terms of two-choice categorization rules help shielding against distraction from irrelevant stimulus attributes. During task switching, this task shielding should temporarily be relaxed in order to prevent the perseveration of the previous task, on the downside making the system more vulnerable towards the intrusion of irrelevant information. Participants had to switch between a digit and a letter categorization task. An irrelevant stimulus feature (Experiment 1: Color, Experiment 2: Font) varied randomly, orthogonal to the task. The presence or absence of an interaction of the irrelevant feature (switch vs. repetition) and the response (switch vs. repetition) was taken as evidence for the absence or presence of task shielding, respectively. Replicating previous results, no feature X response interaction was found on task repetitions, indicating successful task shielding. On task switches however, the feature X response interaction was significant, reflecting the assumed relaxation of task shielding during task switching.
SY_06.5 - Cognitive control and other instructions: An associative learning perspective
Verguts, T.
Ghent University
A first step toward formulating cognitive control in terms of associative learning was taken by Botvinick et al. (2001), who used the Cohen et al. (1990) Stroop model to account for extant cognitive control phenomena such as the Gratton effect. Here, task demand representations bias bottom up task processing, thus overcoming irrelevant sources of information (i.e., cognitive control). However, an explanation of a key aspect of cognitive control is lacking: How are task demand representations implemented on the fly such as to bias task processing? I take the associative learning approach a step further and show how task representations and its biasing of task processors can be conceptualized as fast implementation of instructions, produced by Hebbian learning (Verguts & Notebaert, 2009). This view is instantiated in a computational model. Empirical data consistent with the model are discussed.
SY_06.6 - Instructed and practiced Flanker and Gratton effects
Wenke, D. 1, 2 , Liefooghe, B. 2 & De Houwer, J. 2
1 Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig
2 Ghent University
Functional similarities between practiced and merely instructed S-R mappings were explored using an Eriksen Flanker task. We compared the overall Flanker compatibility effect (FCE) and its sequential modulation (i.e., the Gratton effect) in a practiced condition with an instructed condition. Four new S-R mappings were instructed on each block of trials. In the practiced condition, each instructed stimulus could appear as target and as flanker. In the instructed condition, two stimuli only served as targets, whereas the other two exclusively appeared as flankers. We found significant FCEs in both the instructed and the practiced condition. Moreover, the Gratton effect was similar in the two conditions: It only showed in target/response repetition trials, but not in alternation trials. Our results suggest that instructions alone can induce bindings between S- and R features that lead to automatic response activation when instructed stimuli appear as flankers in a functionally similar way as practiced flankers.