#57758 | AsPredicted

'A1IEToM-BehOnline: Online Adult Study of Altercentric/Egocentric RT Biases'
(AsPredicted #57758)


Author(s)
This pre-registration is currently anonymous to enable blind peer-review.
It has 2 authors.
Pre-registered on
02/08/2021 08:23 AM (PT)

1) Have any data been collected for this study already?
It's complicated. We have already collected some data but explain in Question 8 why readers may consider this a valid pre-registration nevertheless.

2) What's the main question being asked or hypothesis being tested in this study?
The present study addresses the question whether two opposing biases (altercentric and egocentric) are involved in implicit and explicit Theory of Mind (ToM) tasks. To this end, two hypotheses are tested:
H1: Reaction times in an object detection task are modulated by the (task-irrelevant) belief of an agent (altercentric bias, implicit ToM).
H2: Reaction times in an action detection task are modulated by the (task-irrelevant) belief of the participant (egocentric bias, explicit ToM).

3) Describe the key dependent variable(s) specifying how they will be measured.
Button presses are recorded via the Online platform Pavlovia. Only left and right button presses that occur between frame 595 (23.8s after video onset) and frame 660 (26,4s after video onset) of a respective video are included into the analysis (For implicit trials, from when a part of the object becomes visible until a curtain becomes visible that ends the scene. For explicit trials, from when the first eye movement of the agent can be seen that indicates the direction of her action until the curtain becomes visible.). Reaction times (RTs) are calculated relative to the occurrence of frame 595 in each respective video. Depending on the analysis (see 5)) individual reaction times (IRTs) or average reaction times (ARTs, i.e. averaged over button press side) for each subject in each experimental condition will be utilized as dependent variable (DV).

4) How many and which conditions will participants be assigned to?
During the experiment, participants watch 16 short video clips (test videos) in two consecutive experimental blocks. The 16 test videos differ with respect to the following three factors:

I) Belief (B): In half of the test videos, the agent will wear an opaque mask while an object transfers from one box to another and will thus hold a false belief (FB) about the object’s location. In the other half, she will wear a transparent mask while the object switches place and will thus hold a true belief (TB) about the object’s location. Participants are familiarized with both masks at the beginning of the experiment, learn their transparency, and their understanding is checked via control questions.

II) Reality Congruency (RC): In half of the videos, the outcome will be congruent (C) to the real object location. In the other half, the outcome will be incongruent (I) to the real object location.

III) Task (T): During the first half of the test videos, participants complete an object detection task (implicit ToM). During the second half, they complete an action detection task (explicit ToM). A block-wise design is chosen for these conditions to assure that the explicit ToM task does not influence the performance in the implicit ToM task. Each block is preceded by a training that consists of two short training videos and two control questions concerning the transparency of the masks. Training videos are repeated until the participant gives a correct answer at least once to each video type.

5) Specify exactly which analyses you will conduct to examine the main question/hypothesis.
H1: Primary analysis will be done via a generalized Bayesian linear mixed model (GLMM) using default priors. We will include IRTs as DV, B and RC as independent variables (IVs) and a random intercept per subject. We expect an interaction (IA) between B and RC and a main effect (ME) for RC. That is, we expect shorter RTs for C compared to I trials. In addition, we expect that FB trials will be faster than TB trials in I trials (planned comparison, directed Bayesian paired t-test with default priors). For C trials, we expect that TB trials will be faster than FB trials. In a second step, we will include side of the button press and trial order as covariates in the model. There might be a ME for the factor button press side depending on participants handedness. Also, there might be an interference effect of I, i.e. if the first trial is an I trial, participants may be faster on subsequent I trials and slower on subsequent C trials. Also, there might be an interference effect of FB, i.e. if the first trial is a FB trial, participants might be faster on subsequent FB trials and slower on subsequent TB trials.

Secondary analysis will be done with a repeated measures Bayesian ANCOVA using default priors with the within-subject factors B and RC and the between-subject covariates first-belief (i.e. the first trial was a TB trial vs the first trial was a FB trial) and first-congruency (i.e. the first trial was a C trial vs the first trial was an I trial). The DV will be ARTs. Expectations and planned comparison will be analogous to above.

H2: Primary analysis will be done via a Bayesian GLMM using default priors and the same DV and IV as for H1. We expect an IA of B and RC, and a ME for the factor B (i.e., shorter RTs for TB compared to FB trials). More specifically, we expect that I trials will be faster than C trials in FB trials (planned comparison, directed Bayesian paired t-test with default priors). For TB trials, we expect that C trials will be faster than I trials. There could also be an additional ME for the factor C depending on whether the egocentric bias outweighs the agent’s belief. As for H1, we will include side of the button press and trial order as covariates in a second step.

Secondary Analyses: Analogous to H1.

In case, the assumptions for the statistical tests are violated in the data, we will adjust our analyses accordingly.

6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
We will exclude responses that are incorrect, that were given too early (before frame 595+150 ms) or too late (after frame 660), that were affected by any technical errors, that indicate non-meaningful task-compliance or that are more than three standard deviations above or below the participants overall mean.

Participants will be excluded if they show non-meaningful task-compliance (i.e. more than 25% errors in total or more than one incorrect control question per block), if their reaction times differ substantially from the rest of the sample (e.g. very long reaction times due to a slow internet connection etc.) or if they have any history of neurological/psychiatric disorders.

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.

We are going to collect data until the Bayes Factor of the interaction between RC and B of the GLMMs reaches 4 or 0.25. To control false positives and negatives, we will collect a minimum of 68 participants.

8) Anything else you would like to pre-register?
(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)

Data collection has already started (40 participants), but we have not looked at the data at all except to keep track of participation rate.

We will take a look at learning effects across trials and interference effects. In addition, we will examine effects of age, gender and education, compute an item analysis and contrast subjects that were aware of the manipulation in the implicit task with those that were not aware of the manipulation.

In an online study, there is a lot of variance in hardware and software across participants. To assure that any results are not confounded by operation system, frame rate, browser or screen size, all will be added in an additional control step as covariates in the different models mentioned in the analysis section.

Depending on recruitment success we might recruit from different participant pools. In this case, we would include participant pool as a covariate in all analyses.

In parallel, we are conducting a preregistered lab study (AsPredicted #50138, ‘A1IEToM-Beh: Altercentric and Egocentric Reaction Time Biases in Adults'.) using the same stimuli mentioned above. To compare online and lab results, we will perform an additional analysis including only the first individual responses of the lab study and the online study.