'semantic_representation_free_choice'
(AsPredicted #81739)
Download PDF

---

---

1) Have any data been collected for this study already?
No, no data have been collected for this study yet.

2) What's the main question being asked or hypothesis being tested in this study?
The experiment is part of a line of experiments concerned with the ideomotor principle and the acquisition and usage of action effects for action selection. The main question of this experiment is, whether the spatial (top/bottom) positions of visual events (presented as effects) are semantically represented (in an amodal way) or only according to their modal features (i.e., their actual location). To this aim, we build on a study from Hommel, Alonso, and Fuentes (2003, Visual Cognition), where generalization from an exemplar word to a corresponding category has been reported and employ the "induction paradigm" with a free-choice learning phase, but we use a free-choice test phase (as in Elsner & Hommel, 2001, Exp. 2-4). Stimuli in the test phase are the centrally presented German words OBEN and UNTEN ("up" and "down").
• H1: For the control group (effects in the learning phase are the centrally presented words OBEN and UNTEN), there will be a response bias in the test phase towards the respective key associated to a certain effect during the learning phase, which now serves as an imperative stimulus eliciting the response.
• H2a: If generalization occurs, the response bias will be of the same amount for both of the two groups (control group and experimental group).
• H2b: If no generalization occurs, the response bias will be smaller in the experimental group compared with the control group, and within the experimental group perhaps absent at all. A remaining (smaller) response bias in the experimental group indicates that some, but not full, generalization occurred.

3) Describe the key dependent variable(s) specifying how they will be measured.
• The key dependent variable is the response bias during the free-choice test phase. A congruent response is one where the chosen key is the one associated to the stimulus during the learning phase.
• If participants choose randomly without any influence of the learned associations, the frequency of congruent responses should be 50%. Any value larger than 50% of congruent responses indicates a response bias.

4) How many and which conditions will participants be assigned to?
• Participants will be assigned to one of two groups during the learning phase: (1) the experimental group, where a left/right key press elicits the appearance of a visual stimulus in the upper or lower part of the screen, or (2) the control group where a left/right key press elicits the appearance of either the word "OBEN" (German for "up") or "UNTEN" (German for "down") in the screen center.
• There will be four different response-effect-mappings during the learning phase for counterbalancing. The participants will be assigned randomly to one of the four mappings:
 Experimental group: left key -> upper effect, right key -> lower effect
 Experimental group: left key -> lower effect, right key -> upper effect
 Control group: left key -> "OBEN", right key -> "UNTEN"
 Control group: left key -> "UNTEN", right key -> "OBEN"

5) Specify exactly which analyses you will conduct to examine the main question/hypothesis.
• The main analyses will focus on the percentages of congruent response choices in the test phase (response bias).
• To analyze if a response bias is present in one of the groups, a separate Bayesian t-test will be conducted for each group (control group and experimental group) to assess whether the frequency of congruent responses deviates from a value of 50% (see above, point 2).
• To test if generalization is present, a Bayesian two-sample t-test will be conducted with both groups. Generalization can be assumed if the response bias in both groups is of the same size.

6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
• Participants who choose one response key less than 80 times during the 200 learning phase trials will be excluded from further analyses (as in Hommel et al., 2003).
• Participants who respond in more than 20% of the no-go trials (see below, point 7) in the test phase will be excluded (as in Dutzi & Hommel, 2009).

7) How many observations will be collected or what will determine sample size?
No need to justify decision, but be precise about exactly how the number will be determined.
• The sampling plan is to conduct sequential sampling by calculating and monitoring the Bayes Factor, starting from a minimum sample size of n = 20 per between-subject group (control group and experimental group), that is, a minimum number of n = 40 participants in total.
• The Bayes Factor will then be monitored by calculating it each time after the data of four new (counterbalanced) participants was collected.
• We decided to apply the following stopping rules, according to our hypotheses:
Sequential sampling will be stopped when (see Schönbrodt & Wagenmakers, 2018, Schönbrodt et al., 2017)…
a) …a BF10 < 1/10 is calculated for the one-sample t-test of the control group. This result would mean that participants have reacted randomly during the test phase instead of showing a response bias towards either congruent or incongruent responses, suggesting that no learning took place during the acquisition phase of the experiment.
b) …a BF10 > 6 is calculated for the one-sample t-test of the control group, indicating that learning took place in this group, and at the same time either a BF10 < 1/6 or a BF10 > 6 is calculated for the two-sample t-test. The first result would mean that evidence points towards our H2a, the second result that evidence points towards our H2b.
c) In case results are in favor of H2b, as an exploratory analysis, we calculate a (Bayesian) t-test for the experimental group to assess, whether signs of some generalization are observed (BF10 > 1) for this group.
d) A maximum number of n = 50 participants per group has been reached.

8) Anything else you would like to pre-register?
(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)
The test phase employs a free choice task with included no-go trials to prevent participants from preplanning their response before onset of each stimulus (see Elsner & Hommel, 2001, Exp. 3). Taking their Exp. 3A as an example for this experiment, we decided to employ a total of 200 test trials. Thus, in the test phase with 200 trials, the word "OBEN" will be presented in the screen center 50 times and the word "UNTEN" will be presented in the screen center 50 times (50% go trials), while a letter string without semantic value ("XXXXX") will be presented 100 times to indicate no-go trials (50% no-go trials). No-go trials and stimulus trials will be randomly intermixed, so that participants cannot foresee which kind of stimulus will be presented in the next trial.

Literature:
Dutzi, I. & Hommel, B. (2009). The microgenesis of action-effect binding. Psychological Research, 73, 425 – 435.
Elsner, B. & Hommel, B. (2001). Effect anticipation and action control. Journal of Experimental Psychology: Human Perception and Performance, 27, 229 – 240.
Hommel, B., Alonso, D. & Fuentes, L.J. (2003). Acquisition and generalization of action effects. Visual Cognition, 10 (8), 965 – 986.
Schönbrodt, F., Wagenmakers, E.J., Zehetleitner, M. & Perugini, M. (2017). Sequential hypothesis testing with Bayes factors: Efficiently testing mean differences. Psychological Methods, 22(2), 322 - 339.
Schönbrodt, F. & Wagenmakers, E.J. (2018). Bayes factor design analysis: Planning for compelling evidence. Psychonomic Bulletin Review 25, 128 – 142.