Page 13 - dan_sun

Introduction and Overview

ner. This account does not assume the acquisition of rather sophisticated causal knowledge,
nor a decision-maker who deliberates on selecting goals and means. Goal-directed action is
simply believed to result from the ability to form bidirectional associations between actions
and outcomes, which capture a history of cooccurrences of acting and perceiving the result- 1
ing events. Similarly, the initiation of goal-directed behavior would not require any deliber-
ation or even intentions. As the case of mimicry suggests, behaviors could be activated by
merely perceiving their associated outcomes or events.
Outline of the Dissertation

The current dissertation critically examines the evidence for this mechanistic ac-
count for goal-directed behavior. The addressed questions are basically two-fold: First, is
ideomotor learning indeed the result of bidirectional associations formed as a result of re-
peated cooccurrence of actions and outcomes? And second, are ideomotor effects on action
indeed the result of mere activation of the outcome representation.

Chapter 2 examines the evidence for ideomotor action obtained in the literature
featuring the two-stage paradigm inspired by Greenwald's work (1970). The argument starts
that while the learning phase in this paradigm was designed with the mechanistic account
and Hebbian learning in mind (i.e., hundreds of trials in which usually two finger press-
es each produce a specific sensory outcome), this learning phase does not rule out other
forms of learning. That is, every box is ticked to facilitate the formation of bidirectional as-
sociations through mere repetition. However, people can also easily make inferences about
the causal structure of the task (e.g., pressing the left finger causes a high-pitched tone),
based on which the hypotheses or propositions about the relations between actions and
outcomes can be formed. Likewise, although the mechanistic account would indeed predict
that stimuli related to the outcomes featured in the learning phase would bias behavior in
the direction of the associated action, other explanations for these effects can be offered.
Putting together, these arguments provide support for the assumption that the actual evi-
dence for the mechanistic account of ideomotor actions is relatively thin, to say the least,
and that more work is needed to put this account to the test.

Chapter 3 investigates in line with this reasoning to which extent learning in the
two-stage paradigm is spontaneous (Sun et al., 2020). Even though learning should be the
result of mere cooccurrence of actions and outcomes, instructions almost always emphasize
the causal nature of the relation between actions and the events that follow them. Further-
more, two different forms of testing for ideomotor effects are used: the Free Choice Task,
in which people can freely choose which action to perform, and the Forced Choice Task, in
which people have to react to an imperative stimulus with an instructed response.

While both tasks are used frequently in the literature, only Experiment 1 featuring
the Free Choice Task produced the ideomotor effect. That is, when stimuli that previously
served as outcomes in the acquisition phase were presented together with the imperative
stimuli in the Free Choice Task, it could bias action towards the corresponding responses.
This effect occurred regardless of whether the instructions emphasized the causal nature

11