What determines visual search ability?

Background

Imagine you are playing a first-person shooting game (e.g., though it's overly simplified, imagine the game requires you to shoot the red T indicated in the white circle among skewed Ls), how fast and accurately are you able to find your shooting target? What are the underlying mechanisms that enable you to visually search for it? Can we evaluate your search ability through some other tests?

Research Question

What are the visual abilities involved in your behaviors of looking for something (i.e. visual search), and to what extent is each visual ability contributing to how fast and accurate you find the target?

Our Strategy

Step 1: Disentangle the search process and identify visual abilities that might be involved:

Possibility 1: Visual Working Memory Capacity

Your ability to hold a target image in mind so that you can use it for search.

Possibility 2: Object Recognition Ability

Your ability to identify whether an item is identical to the target you hold in mind.

Step 2: Measure visual search performance, visual working memory capacity, and object recognition ability. Test whether the last two predict visual search performance.

Operationalization

How is each visual ability measured?

Visual Working Memory Capacity: How many visual items can you hold simultaneously in mind?

Object Recognition Ability: Can you identify a complex visual object from others that are very similar?

Visual Search Ability: When you are looking for something, how fast and accurately can you find it?


Study Preparation

Visual search and visual working memory test programmed in MATLAB.

Novel Object Memory Test (NOMT) programmed in JavaScript and provided by the authors who developed the test.

Sample

187 participants took part in the study

Experimentation

Task 1: Visual Search


Search for target (the red T in the white circle) among distractors of medium (medium search) or high (difficult search) similarity to the target.


Key measurements: search accuracy and efficiency.

Task 2: Visual Working Memory Test


Remember all the colors shown in a "memory set" and respond whether the color shown later belongs to the previous "memory set".


Key measurement: the maximum number of colors in the "memory set" that elicits accurate response.

Task 3: Novel Object Memory Test


Learn novel visual objects and identify the correct object from similar ones.


Key measurement: accuracy in test questions.

Analysis: Structural Equation Modeling

Principal Component Analysis

Identified 2 latent factors that capture over 50% of the variabilities among individual performance of the 3 tasks.

Exploratory Factor Analysis

Determined to what extent is each measure contributing to the latent factors.

Named the 2 factors as Cautiousness and Cognitive Ability.

Confirmatory Factor Analysis

Developed the overall model.

Results

People differ in their cognitive ability which explains why some people search faster than others.

Cognitive ability also predicts how well people identify complex objects, and how many objects people can simultaneously hold in mind to perform cognitive tasks (e.g., for identifying whether a new object is already seen).

People's cognitive ability then predicts how cautious they are in visual search, reflected in how fast they proceed with difficult search situations and how accurately they search.

Confirmatory Factor Analysis results shown below.

Implications

In activities involving visual search (e.g., locating a shooting target in video games), a person's visual working memory capacity and ability to recognize complex objects can predict how fast and accurate they search.

Therefore, to evaluate a person's visual search ability (e.g., identify competent video game players), we can also use a visual working memory test and the object recognition test (NOMT) as measurements.

Delivery

This project was presented at a seminar talk at Psychology Department at University of Illinois at Urbana-Champaign.

This series of experiments was also presented as posters at Vision Science Society 2018 and 2019 Annual Meetings.