| Popis: |
This is the second of a series of experiments on time-based expectancy for finger responses to vibrotactile stimulation. In this experiment, we induce time-based expectancy by pairing two events and two preparatory intervals. In a binary randomized response choice setting, participants respond by keypresses to vibrotactile stimulation to their fingers. Stimulus response mapping is compatible (i.e. respond with the finger that was stimulated), and stimuli are preceded by either a short or a long preparatory interval which predicts with 90% probability which finger will be stimulated next. This pairing of vibrotactile stimulation (left-right finger) and interval duration (short-long) is counterbalanced across participants. For half of the participants the short interval will predict stimulation on the left finger, and the long interval will predict stimulation on the right finger with 90% validity. For the other half, this relation will be inverted. We decide between two hypotheses: 1) Participants respond about equally fast to frequent combinations of interval duration and stimulation side than infrequent ones. Such finding would mean that participants did not form time-based expectancies to vibrotactile stimuli. 2) Participants respond faster to frequent than infrequent combinations of interval duration and stimulation side. Such finding would mean that participants build time-based expectancies to vibrotactile stimuli, and would be consistent with the time-based expectancy literature (see, Thomaschke et al., 2011; Thomaschke et al., 2015) We decide between these hypotheses by a Bayesian analogue to a paired sample t-test (Morey & Rouder, 2011). We cumulatively test participants until we reach a JZS Bayes factor over 5 for either of the hypotheses, or until we have tested 120 participants without reaching a JZS Bayes factor over 5. We collect data with the E-Prime program uploaded to this repository. We will pre-process the data as follows: We do not include participants with an overall mean RT or an overall mean error rate lying more than 2.5 SDs away from the sample mean. We do not include the first trial of each block, the first block, error trials, and trials following an error. Since we will compare two conditions observations (frequent vs. infrequent) with different number of, we will follow special recommendations to calculate the median RT. We will take the regular RT median for the infrequent condition and we will compare it with a special “average median” from the frequent condition, separately for the short and long intervals durations (for a more detailed explanation, see Miller, 1988) By frequent we mean trials with combinations of interval duration and stimulation side with 90% validity and as infrequent count trials with combinations of interval duration and stimulation side with 10% validity. We conduct the Bayesian t-test over these conditions. All further analyses will be of exploratory nature: For example, effects on error rates, effects of interval duration, etc. For following exploratory analyses, the script will be extended. |
