Posters: Implicit Cognition

We affect our environment through our actions. In doing so, we generally feel in control of our actions and their consequences, a phenomenon which has been termed sense of agency. Agency is well documented for actions that aim at causing perceivable effects in the environment, but not all actions aim at causing an effect – rather, actions may also aim at preventing an event which would occur otherwise. Such prevention behaviour poses a critical challenge to the cognitive system, because successful prevention inherently revolves around the absence of a perceivable change. I will present a series of experiments showing that this state of affairs leads to a profound dissociation of explicit and implicit measurements of agency: Whereas participants reported high levels of agency in explicit judgements, there was no sign of agency in implicit measures of temporal binding. These results attest to an altered action representation for prevention behaviour, in line with current theories of clinical avoidance learning, and they might help to better understand avoidance behaviour.

In spite of the common use of green and red as signals for truth and falsity in everyday life (e.g., in quizzes or exam marking), the possible link between these colours and concepts lacks systematic investigation. For this reason, we used a stroop-task to test the hypothesized associations between green and truth as well as red and falsity. Words semantically related to truth and falsity appeared in red, green or grey on the computer screen. The participants’ task was to categorize each word based on its semantic meaning (i.e., true or false) and to do so as fast and accurately as possible. The results yielded strong evidence for the predicted associations: Reaction times primarily revealed facilitation effects in the form of faster categorization of truth-related words presented in green than in the other colours while falsity-related words were categorized faster when presented in red. In contrast, error rates mainly showed interference by incongruent colours (e.g., truth-related words presented in red). To our knowledge, this is the first time that automatic associations between green and truth as well as red and falsity have been demonstrated by means of an implicit measure.

In a previous study, we found that a sequence of task-irrelevant effect-tones can be learned implicitly when the effects are not consistently mapped to the responses. Furthermore, this effect-sequence can trigger the development of explicit knowledge about a later induced contingent response-sequence. However, with visual effects, we were not able to replicate these findings (Lustig & Haider, 2019). One possible explanation is that visual effects are generally less salient than auditory effects (Ziessler & Nattkemper, 2002). Alternatively, the visual effects might have interfered with the visual presented targets as both are presented in the visual modality (e.g. Mayr, 1996; Ruess et al., 2018). To investigate this issue, participants performed a serial reaction time task by responding verbally to randomly presented auditory stimuli. The experimental condition received visual presented response effects, whereas in the control condition tones were presented as response effects. In both conditions, the respective effects were contingently mapped to the verbal responses and both followed a correlated sequence. First results indicate that the participants in the experimental condition had more explicit knowledge about the response sequence than the participants in the control condition. We conclude that the modality of the response effects and the to-be-performed task mediates the influence of response-effects on the development of explicit knowledge about a response sequence.

When we are to judge how well we will perform in an upcoming task, we might use knowledge about comparable past performances. Yet, objective and reliable performance measures are not always available and thus it might be helpful to base a performance prediction on past metacognitive judgments which can be independent of actual performance. In our study, we used a metacontrast priming task in which the time between the prime and the mask was varied (SOA). This manipulation has been shown to lead to situations where the objective performance is equal while the subjective feeling of correctness can vary (Lau & Passingham, 2006). Additionally, we paired the different SOAs with a preceding 100% predictive cue. In a subsequent test task, we found that the presentation of these cues led to a prediction of performance based on previous metacognitive judgments which can be independent of objective past performance.

The Implicit Association Test (IAT, Greenwald, McGhee & Schwartz, 1998) has been used in almost every field of psychology. Despite its wide spread application as an implicit measure of attitude, the IAT was criticized for different methodical reasons (e.g. confounding with block order, possible recoding of the task, voluntary controllability). Various studies found effects of block order, stimulus material or training in the classical IAT, questioning the interpretation of the IAT effects as solely measure of implicit attitudes. However, no study has yet investigated at the same time different effects of the IAT setup per se. Here, we used the typical insect-flower-IAT in a well-powered study (N = 90) to systematically test for the effects of block order (insect vs. flower combined with positive first) and valence category assignment (positive in the left corner vs. negative in the left corner) as between subject variables. Stimulus type (pictures or letters), stimulus category (positive, negative, insects or flowers) and block type (crossed [e.g. insect + positive] or recrossed [e.g. insect + negative]) were within-subject variables. It was shown that the IAT effect varied based on stimulus type, stimulus category and block order. Further, reaction times for the four stimulus categories differed based on the stimulus type. Another significant interaction was found between stimulus type and block order. Taken together, our research shows that various factors in the IAT setup influenced the underlying IAT effect limiting its validity as an implicit measure for associative preferences.

Evidence suggests that the ability of the implicit association test (IAT) to predict behavior is rather low although numerous moderators have been identified. In order to tackle this issue we want to introduce a so far overlooked moderator to the IAT, that is the test difficulty from classical test theory. It is long known that the relationship between two tests depends on their test difficulty (congruency between test difficulty leads to higher correlations than incongruency). Thus, the aim of our first study was to develop IATs with different difficulties and to test the proposed correlational pattern. In total 97 participants took part in our study which consisted of a 2x3 within design with the factors measurement (IAT and behavior) and difficulty (easy, moderate and difficult). We were able to manipulate the difficulty of the IATs, however the expected correlational pattern failed to appear. One possible reason for this unexpected finding is the fact that the moderating effect of test difficulty has been argued to result from the distribution of the data. And indeed the distribution of the individual IAT-effects was close to normality regardless of the difficulty of the IAT whereas the distribution of the behavior measures showed expected ceiling and floor effects. In a second study we thus want to test whether we can induce such ceiling and floor effects into the IAT experimentally by manipulating the attitude towards a priori neutral objects and developing IATs with extreme difficulties. Results and implications for IAT-research will be discussed.