Using ACT-R and SGOMS to Predict Micro Strategies Used by Experts During Routine Tasks

Abstract

Micro strategies refer to fast, low level, unconscious strategies that constitute a control structure for information processing within a specific task (Newell, 1973). Micro strategies determine how we process low level information, such as perceptual inputs (Shiffrin & Cousineau, 2004) or procedural information (Gray & Boehm-Davis, 2000). Using an SGOMS/ACT-R model I was able to predict expert, individual performance on a stimulus-response reaction time game involving a memorized hierarchical structure. I found that slightly different micro strategies were used by the participants, indicating that micro strategies have a very real influence on individual performance. These findings represent an understudied area of research that has larger implications for many Psychology and Cognitive Science studies. 2Chapter: The Game……………………………………………………………….……..3 2.1The Methods………………………………………………………………………4 2.2The Unit Tasks…………………………………………………………………….4 2.3The Planning Units………………………………………………………………..5 2.4The Conditions…..………………………………………………………………..6 2.5 App/ Game Details………………………………………………………………..7 3 Chapter: Methodology………………………………………………………………….10 3.1 Data Cleaning…………………………………………………………………….11 4 Chapter: The Model……………………………………………………………………11 4.1 Code Examples for each Condition……………………………………………...13

Understanding Micro Cognition using SGOMS/ACT-R predictions of a SGOMS based Mobile application

“…The game in Study 1 was based on an SGOMS/ACT-R model (Greve, 2021) developed based on a previous pilot study. The new model assumed buffer storage for the planning unit existed and that planning unit information was updated and held in this buffer throughout planning unit execution.…”

“…For a complete account of the methodology used in this study, see Greve (2021) and Greve, Reid & West (2020). Essentially, subjects practiced the game extensively until they attained an expert level of play as measured by reaction time.…”

“…The author of this study was involved in an early publication (Greve et al, 2020) and programmed the game used in Study 1. Since Study 1 has already been published (Greve, 2021), this thesis focuses on Study 2. However, since Study 2 is a follow-up on Study 1, it is essential to understand the results of Study 1 in detail.…”

“…To do this we studied the micro strategies used by participants when playing an SGOMS based game. This research explores micro strategies by comparing two computational SGOMS/ACT-R models to predict participant performance under different conditions (for previous research on this, see: Dudzik et al, 2018;Greve et al, 2021;Greve et al, 2020).…”