Using Brain Computer Interfaces to Understand Distractions During Virtual Reality Tasks

Attentional control is essential in order to focus on relevant information and fade out distracting events. The consequences range from reducing the enjoyment and quality of life (e.g., not being able to focus while reading a good book) to affecting the ability to concentrate at work or even causing accidents (e.g., while driving).

Research has suggested that the extent to which irrelevant or distracting events are perceived depends on the perceptual load of a given task (Lavie et al., 1997). If attentional resources are exhausted (high perceptual load), there is little capacity left to process irrelevant information. However, if perceptual load is low, attentional resources can easily spill over to distracting events which interfere with task-relevant information. This phenomenon has been shown in a variety of different contexts such as mind wandering, which can be reduced if perceptual load is high. For patients with attention deficit disorder (ADHD) it has been suggested that distractibility can be reduced if perceptual load is high.

In this research we aim to use functional Near Infrared Spectroscopy (fNIRS) methodology to understand how perceptual load manipulation in VR tasks affect the attentional resources in the face of irrelevant distractions.

This PhD is structured around the 4 work packages (WP) presented below:

WP1: Baseline Data Collection using fNIRS during VR perceptual load tasks.
This first phase of this work will be focused on replicating and extending existing research started by Lavie et al., 1997 which showed, using a motion pattern of moving dots in the periphery of the participant view, that the task load and operator workload affects the perception of irrelevant distractors in the environment. Compared to the original work, this research will use non-invasive brain sensors including fNIRS to evaluate participants' mental workload during tasks designed in VR. We will manipulate and control the demands placed upon participants by designing different task load conditions (High vs Low load). The critical test of our hypothesis is whether evoked activity related to irrelevant visual motion (distractors) is smaller under conditions of high processing load (compared with low load).

WP2: Exploring techniques from Data Science for advanced analysis and classification of user state based on brain activity data.
While this WP is not directly contributing to the Data Science research with new algorithms, but will focus on using existing Data Science techniques and classification methodologies to design a working pipeline and framework for analysing fNIRS data in the context of user state detection. A good starting point is the work presented by Benerradi et al. 2019, who investigated the use of three different approaches: a logistic regression method, asupervised shallow method (SVM), and a supervised deep learning method (CNN), to classify users' workload during tasks.

WP3: From controlled to more realistic and complex task scenarios.
Starting in controlled settings with baseline data from fNIRS during simple memory tasks-WP1, this work will move towards evaluating more complex and realistic tasks (WP3). Classroom environments, driving a car or video games can be applied in VR in order to create more realistic environments in which we often get distracted even though concentration is required. Those environments can be transformed as training tools in order to increase our ability to ignore distractions (Anguera et al., 2013; Gazzaley & Rosen, 2016).

WP4: Design Framework based on the WP1 and WP3 results.
This framework will be based on the results from WP1 and WP3 and will further contribute to the work on attention and distractions presented by Gazzaley,A., and Larry,D.R. (2016) with guidelines for designers of future technologies, as well as researchers in the field of HCI and Psychology.