Ziel

Abstract (2018-2021):
While previous research on human task switching has accumulated convincing evidence for the existence of task representations in the human mind, studies concerning the implicit and explicit learning of these task sets are sparse (but see Dreisbach, Goschke, & Haider, 2007). Typically, studies on task switching comprise supervised experimental procedures, within which the relevant control policies (or abstract rules) are explicitly known to the participants and refer to already established stimulus-response mappings. In the literature on sensorimotor learning, however, one can find more than a hundred years of research investigating the acquisition of novel, unknown control policies by means of supervised, unsupervised and reinforcement learning (e.g. Galea, Mallia, Rothwell, & Diedrichsen, 2015; Heuer, Hegele, & Suelzenbrueck, 2011; Todorov, 2003; Welch, 1972; Helmholtz, 1909; Stratton, 1896). Most of this research, however, has focused on the acquisition of a single a control policy as for example in learning how to move under a novel sensorimotor transformation introduced by prism goggles, force-fields or visuomotor rotations. Among those who sought to investigate simultaneous learning of multiple control policies, only very few have addressed the simultaneous learning at different levels of processing, i.e. the interplay of (implicit) learning of internal representations of abstract rules based on the experienced stimulus-response- mappings and (explicitly and/or implicitly represented) states that contextualize those abstract rules and can potentially give rise to (explicit) cognitive control of otherwise spontaneously executed movements. Within the SPP, the present project offers a unique perspective on task-switching in terms of the way humans simultaneously acquire multiple control policies for goal-directed behavior.
Abstract (2015-2018):
The project investigates simultaneous learning of multiple sensorimotor transformations. Previous research on this topic revealed inconclusive evidence regarding the concurrent development of two internal models of the transformations. Based on unresolved issues in sensorimotor learning, the project will [a] shed light on the nature of contextual cues that allow for the simultaneous development of two internal models in dual adaptation, [b] examine the way in which explicit knowledge of the transformations and its application in terms of conscious strategic corrections modulate the implicit acquisition of multiple internal models of the transformations, and [c] systematically study the influence of the type of transformations to be learned.
To this end, the project will investigate dual sensorimotor learning by breaking it down into its explicit and implicit component processes by a series of pre- and posttests (Heuer and Hegele, 2008). Three hypotheses constitute the theoretical basis for the planned series of experiments.
(1) The distinct-motor-requirements hypothesis postulates that the amount of interference between any novel transformations depends on the extent to which there is a conflict between the adjustments required to the adapted motor commands;
(2) The strategy hypothesis proposes that learning of multiple visuomotor transformations is mediated by explicit knowledge of the transformations;
(3) The vector-adaptation-hypothesis states that adaptation of movement amplitude and movement direction is fundamentally different phenomena with distinct neural substrates and different learning characteristics such as the time course of learning or the pattern of generalization.
In three series of experiments, we will examine 1) the roles of different contextual cues, 2) the interaction of explicit and implicit components 3) the impact of different types of transformations on subjects’ ability to concurrently learn two visuomotor transformations (dual learning). In addition to the individual knowledge gains by these three approaches, we aim to generate general theoretical insights into the roles of explicit and implicit mechanisms in performing and learning two tasks at the same time.

Zusammenfassung

Der technologische Fortschritt führt zu einer kontinuierlichen Weiterentwicklung von modernen Werkzeugen, die ganz neuartige Anforderungen an die sensomotorische Leistungsfähigkeit von Menschen an ihren Arbeitsplätzen, aber auch im öffentlichen und privaten Raum. Von besonderem Interesse sind dabei Bewegungen, die durch eine räumliche Trennung und mechanische Entkopplung von Bewegungsausführung und- beobachtung gekennzeichnet sind. Diese neue Klasse von Bewegungen finden sich typischerweise in elektronischen Werkzeugen. Ein prototypisches Beispiel ist die Computermaus. In diesem Werkzeug überführt ein elektronisches Gerät die Bewegung der Hand in eine Cursorbewegung auf einem Computermonitor. Im alltäglichen Leben treffen Menschen auf viele unterschiedliche Transformationen dieser Art, deren spezifische Charakteristika sich jedoch stark unterscheiden können und meist von Grund auf im Zuge zielgerichteten Übens neu gelernt werden müssen. Einmal gelernt wechseln Menschen mühelos zwischen unterschiedlichen Werkzeugtransformationen hin und her, beispielsweise beim Fahren eines Autos, wenn ein erfahrender Fahrer problemlos vom Einlegen eines Ganges zum Drehen des Lenkrads wechselt. Die Nutzung moderner elektronischer Werkzeuge erfordert also ein anpassungsfähiges System, das höchst flexibel und in der Lage ist, verschiedenste Werkzeugtransformationen gleichzeitig zu speichern und störungsfrei abzurufen. Ziel des vorliegenden Projekts ist es zu verstehen, wie Menschen mehrere Transformationen gleichzeitig lernen können. Dazu erforsche ich die diesem Multilernen zugrundeliegenden expliziten und impliziten Lernprozesse. Genauer gesagt strebe ich in vier Arbeitspaketen danach zu verstehen, (a) wie die zentralnervöse Verarbeitung unterschiedlicher Fehlerarten explizite und implizite Lernprozesse antreibt, (b) welche Rolle das Kleinhirn, das traditionellerweise mit impliziten Lernprozessen in Verbindung gebracht wurde, für explizite Lernprozesse spielt, (c) ob und wie zusätzliche kognitive Aufgaben das gleichzeitige explizite und implizite Lernen mehrerer Werkzeugtransformationen beeinflusst, und letztlich, (d) ob sich allgemeine Prinzipien der Informationsverarbeitung beim Multilernen finden lassen, anhand derer die sich Grenze der am SPP beteiligten Disziplinen "Bewegungswissenschaft" und Kognitionspsychologie überwinden lassen um so einen Beitrag zum übergeordneten Ziel des Projekts, der Formulierung einer neuen, umfassenderen Theorie menschlichen Multitaskings (in meinem Projekt genauer gesagt: des menschlichen Multilernens), zu leisten.
(DFG- Projektnummer 274922074)

Zusammenfassung

Technological advances continuously lead to the development of modern tools that pose new demands on the sensorimotor capabilities of humans in work settings as well as in public and private spaces. Of particular interest, at least from a perspective of sensorimotor learning, is a novel class of movements that are characterized by the spatial separation and mechanical decoupling of movement execution and observation. These movements are typically found in the use of electronic tools. A prototypical exemplar is the computer mouse. With this tool, an electronic device mediates the transformation of hand movements into motions of the cursor on the computer monitor. In everyday life, humans encounter a wide variety of such electronic tools, whose characteristics strongly differ and most of the time, must be learned from scratch in the course of deliberate practice. Once learned, humans effortlessly switch between various tool transformations, for example, in driving a car, when the experienced driver is able to instantaneously switch from engaging a gear in the horizontal plane to turning the steering wheel in the frontoparallel plane. Thus, the use of modern tools requires an adaptive system that is highly flexible and able to represent multiple tool transformations at the same time. In the present project, I seek to understand how humans can learn multiple tool transformations at the same time. To this end, I investigate explicit and implicit components of human multilearning. In four work packages, I seek to understand (a) how the processing of different kinds of performance errors by the central nervous system drives explicit and implicit learning, (b) the role of the cerebellum, which has traditionally been associated with implicit learning, for explicit, strategy-driven learning, (c) if and how additional cognitive tasks affect explicit and implicit components of learning multiple transformations at the same time, and finally, (d) if it is possible to identify common principles of information processing in human multilearning, with which we can bridge the gap between cognitive psychology and movement science in order to contribute to the overarching goal of the project, the formulation of a novel, more comprehensive theory of human multitasking (in my project: human multilearning).
(DFG- Project number 274922074)