JMIR Publications

ABSTRACT

Background:

Visual search declines with aging, dementia and brain injury and has been linked to limitations in everyday activities. Recent studies suggest that visual search can be improved with practice using computerized visual search tasks and puzzle video games. For use in assessment and intervention, it is important that tools provide difficulty levels so that patients’ visual search ability can be assessed and practiced in a controlled and adaptive way. However, commercial puzzle video games make it hard to control task difficulty and there are little means to collect performance data.

Objective:

The aim of this study was to develop and initially validate the Search & Match Task (SMT) that combines an enjoyable tile-matching match-three (TMM3) puzzle video game with features of the visual search paradigm (‘taskified game’). The SMT was designed as a single-target visual search task that allows control over task difficulty variables and collection of performance data.

Methods:

The SMT is played on a grid-based (width x height) puzzle board, filled with different types of colored polygons (tiles). A wide range of difficulty levels was generated by combinations of three task variables over a range from 4 to 8: height and width of the puzzle board (set size), and the numbers of tile types (distractor heterogeneity). For each difficulty level, large numbers of playable trials were pre-generated using Python. Each trial consists of four consecutive puzzle boards where the goal of the task is to find a target tile configuration (‘search’) on the puzzle board and swap two adjacent tiles to create a line of three identical tiles (‘match’). For each puzzle board, there is exactly one possible match (‘single target’ search). In a user study with 6 healthy young (20 to 28 years), 6 old (64 to 72 years) and 6 oldest old (88 to 94 years) adults played the long (young and old adults) or short version (oldest old adults) of the difficulty levels of the SMT. Participants completed neurocognitive tests that measure cognitive domains engaged by the puzzle game and general cognitive ability.

Results:

Results from the user study indicate that the time to find a ‘match’ is associated with set size, distractor heterogeneity and age group. Results further indicate that search performance is associated with general cognitive ability, selective and divided attention, visual search, and visuospatial and pattern recognition ability.

Conclusions:

Overall, this study shows that an everyday puzzle game-based task can be experimentally controlled and permits data collection to assess visual search and cognitive abilities. Further research is needed to evaluate the potential of the SMT game to assess and practice visual search ability in an enjoyable and adaptive way.