Vorhersagebasierte Mechanismen der Separierung von Repräsentationen in sequentiellen Handlungen bei Mehrfachaufgaben

Verantwortlich

Prof. Dr. Hilde Haider hilde.haider[uk]uni-koeln.de

Weiter verantwortlich

Zeitraum

01.10.2018-30.09.2021

Förderung

Beschreibung

Many everyday situations afford to simultaneously execute two or more tasks consisting of inherent action sequences (e.g., cooking). This sequencing has drawn relatively little attention in the literature on multitasking (Botvinick & Bylsma, 2005; Schiffer, Waszak & Yeung, 2015). Cognitive-control models of dual task performance postulate a strategic bottleneck which serves to hinder crosstalk between the two simultaneously conducted tasks (e.g., Hazeltine et al., 2006; Logan & Gordon 2001; Meyer & Kieras, 1997; Miller, Ulrich & Rolke, 2009; Salvucci & Taatgen, 2008; Tombu & Jolicoeur, 2003). They thus could easily incorporate findings showing that such inherent sequences alter dual task costs, while for structural bottleneck theories it could be difficult to reconcile such findings. Our findings of the first funding period have shown that temporal separation reduces the detrimental effects of dual tasking on sequence learning. We further specified the Schmidtke and Heuer (1997)-thesis that the randomness of the sequence of stimuli and responses in the task paired with the Serial Reaction Time Task (SRTT) disrupts sequence learning. Our results (i.e., Röttger et al., 2017) and discussions among movement science and cognitive psychology groups in the SPP1772 (cf. Broeker et al. 2017) suggest that participants automatically predict (in line with predictive coding) upcoming events and can fail to respect task boundaries when doing so. We assume that across-task predictions involving a task with random stimulus sequence disturbs sequence learning and performance in dual tasking. A pilot study shows first evidence for across-task prediction based on crosstalk between (a) the current stimulus in the task with random sequence and (b) the stimuli due in the next two trials in the SRTT. The first goal (Q1 series) therefore is to provide further evidence for the role of across-task predictions in dual-tasking. In the Q2 series, we focus on the possibility of content-dependent separation of the two tasks, as this should allow keeping the two concurrently presented task-sets separate to foster parallel processing and by this address the central problem in multitasking (Hazeltine & Schumacher, 2016). While the reduction of dual-task costs by temporal separation is predicted by either structural or strategic bottleneck models of multitasking, content-dependent separation is difficult to reconcile with the first class of models. Building on characteristics of everyday tasks, we will ask whether and how task separation is enhanced by: (Q2-A) separable outcomes of action effects, (Q2-B) superordinate goals and natural sequences or (Q2-C) conflicts. This can help to better understand why dual-task costs arise and to link findings from cognitive experiments with the motor research groups of SPP1772 who also discuss practice-induced separation of processing into modular structures, laying the ground for integrating these lines of research.

Description

Many everyday situations afford to simultaneously execute two or more tasks consisting of inherent action sequences (e.g., cooking). This sequencing has drawn relatively little attention in the literature on multitasking (Botvinick & Bylsma, 2005; Schiffer, Waszak & Yeung, 2015). Cognitive-control models of dual task performance postulate a strategic bottleneck which serves to hinder crosstalk between the two simultaneously conducted tasks (e.g., Hazeltine et al., 2006; Logan & Gordon 2001; Meyer & Kieras, 1997; Miller, Ulrich & Rolke, 2009; Salvucci & Taatgen, 2008; Tombu & Jolicoeur, 2003). They thus could easily incorporate findings showing that such inherent sequences alter dual task costs, while for structural bottleneck theories it could be difficult to reconcile such findings. Our findings of the first funding period have shown that temporal separation reduces the detrimental effects of dual tasking on sequence learning. We further specified the Schmidtke and Heuer (1997)-thesis that the randomness of the sequence of stimuli and responses in the task paired with the Serial Reaction Time Task (SRTT) disrupts sequence learning. Our results (i.e., Röttger et al., 2017) and discussions among movement science and cognitive psychology groups in the SPP1772 (cf. Broeker et al. 2017) suggest that participants automatically predict (in line with predictive coding) upcoming events and can fail to respect task boundaries when doing so. We assume that across-task predictions involving a task with random stimulus sequence disturbs sequence learning and performance in dual tasking. A pilot study shows first evidence for across-task prediction based on crosstalk between (a) the current stimulus in the task with random sequence and (b) the stimuli due in the next two trials in the SRTT. The first goal (Q1 series) therefore is to provide further evidence for the role of across-task predictions in dual-tasking. In the Q2 series, we focus on the possibility of content-dependent separation of the two tasks, as this should allow keeping the two concurrently presented task-sets separate to foster parallel processing and by this address the central problem in multitasking (Hazeltine & Schumacher, 2016). While the reduction of dual-task costs by temporal separation is predicted by either structural or strategic bottleneck models of multitasking, content-dependent separation is difficult to reconcile with the first class of models. Building on characteristics of everyday tasks, we will ask whether and how task separation is enhanced by: (Q2-A) separable outcomes of action effects, (Q2-B) superordinate goals and natural sequences or (Q2-C) conflicts. This can help to better understand why dual-task costs arise and to link findings from cognitive experiments with the motor research groups of SPP1772 who also discuss practice-induced separation of processing into modular structures, laying the ground for integrating these lines of research.