This repository contains the experimental data from Patrick J. Rice's DRI (Delayed Rule Inference) experiment, together with the cognitive model designed to reproduce and explain the results.
In the DRI study, participants were asked to perform a simple version of RITL paradigm, in which they were first given a specific rule ("Odd / Index") and then asked to respond to the parity of an input number ("7", which is Odd, and therefore requires a response with the Index finger).
Each trial's rule could be of one of two types:
- Concrete rules, like "Odd / Index", which directl specify which finger (index or middle) to use to respond;
- Abstract rules, like "Odd / A", which require to map the location (left or right) a visual response cue ("A" or "B")) on the screen with the corresponding finger (index or middle).
In addition, the condition called for two different possible applications of the rule:
- Instructed application, where the parity of the number is the same as specified in the instructions (i.e., "7" for "Odd / Index")
- Inferred application, where the parity of the number is different, and therefore the response was not specified in the instructions (i.e., "8" for "Odd / Index", which requires responing with the middle finger).
During the experiment, participants received high-frequency TMS (5 Pulses @ 10 Hz) on the dorsal premotor cortex (PMd) or on the vertex (as a control condition). Stimulation could be adminsitered during the encoding phase.
The results showed a very specific effect, with TMS causing a significant delay in response times (~350ms) only when applied to PMd during the execution phase of abstract rules with inferred application.
No other effect was significant (see Data).
The model was implemented in the ACT-R Cognitive architecture (http://act-r.psy.cmu.edu/), version 7.5.
The model follows a simpe strategy. The timeline of the task can be divided into three processing phases:
-
The Encoding phase_. This phase spans the moment from the onset of instructions on the screen to the moment the participant presses the button. During this phase, the instructions are committed to WM in a rather amodal way, e.g. as a verbal code ("Even, Index").
-
The Preparation phase. During the preparation phase, the verbal instructions are used to prepare mental sensorimotor plans. The translation only happens for the specific motor command specified by the instructions. Thus, "Even, Index" is translated as an activation of an "index" motor chunk, which is placed in WM instead of the word "index".
-
The Execution phase. During the execution phase, the parity is checked. Two things can happen:
3.1 If the parity is verified, then the prepared visuo/motor response is executed as planned.
3.2 If the parity is violated (inferred instructions), then a new preparation phase is initiated. The new plan is then executed.
The effect of TMS occurs only during the Preparation phase for symbolic rules. Thus, it does not show up during the Encoding phase, and shows up instead for the Execution phase only whenever the preparation needs to be re-done (inferred trials).