Dynamic Difficulty Adjustment in a P300 Speller Task for Attention Training

Not Applicable

Completed

• Conditions: Attention
• Interventions: Other: Random task difficulty; Other: Task difficulty adaptation according to iterative learning controller; Other: Task difficulty adaptation according to Arvaneh et al. (2019)

• Registration Number: NCT05576649
• Lead Sponsor: National University of Ireland, Maynooth

Brief Summary

This study compares different adaptation approaches in a P300-based BCI neurofeedback training to improve short-term attention. Participants will spell several words on a computer by focusing on a letter on the screen while their brain activity is recorded (P300 speller). The difficulty of this task will be adapted to train their attention. Their cognitive abilities will be evaluated using a random dot kinematogram, where they are asked to indicate the direction a fraction of moving dots on the computer screen is going in, before and after the training to analyse any improvements. Questionnaires will be used to compare the perceived task load of the different adaptation approaches.

Detailed Description

Healthy participants will complete several runs in a P300-speller task, where the number of flashes per row and column is progressively adapted to their performance (in control group 1 and experimental group) or chosen randomly (in control group 2). Decreasing the number of flashes per row and column makes it harder for the computer to identify the letter the participant is focusing on, which encourages them to improve their focus to maintain their performance. The number of flashes can therefore be interpreted as the task difficulty of the P300 speller task.

The participants will complete a continuous version of the random dot kinematogram, where a fraction of incoherently moving dots will start moving coherently. The participants are asked to indicate which direction these dots are moving in. This task is completed before and after the P300 speller training, so that response times in the task can be compared.

To evaluate the perceived task load of the different adaptation approaches, the participants will complete a questionnaire about fatigue and boredom before and after the training session, as well as the NASA Task Load Index at the end of the experiment.

Study Timeline

• Study First Submitted: 2022-09-28
• Study Start: 2022-10-03
• Study First Submitted QC: 2022-10-09
• Study First Posted: 2022-10-12
• Status Verified: 2023-03
• Primary Completion: 2023-03-13
• Study Completion: 2023-03-13
• Last Update Submitted: 2023-03-21
• Last Update Posted: 2023-03-22

Recruitment & Eligibility

• Status: COMPLETED
• Sex: All
• Target Recruitment: 51

Inclusion Criteria

• 18 years and older
• able to consent
• normal or corrected-to-normal vision

Exclusion Criteria

• history of neurological or cognitive illnesses
• negative reaction to electroconductive gel used in the study (an allergy patch test will be carried out before the experiment starts)
• illiteracy

Study & Design

• Study Type: INTERVENTIONAL
• Study Design: PARALLEL

Arm && Interventions

Group	Intervention	Description
Control Group 2	Random task difficulty	In this group, the number of flashes per row and column during the P300 speller training will be chosen randomly. It is independent of the participants' performance.
Experimental Group	Task difficulty adaptation according to iterative learning controller	In this group, the number of flashes per row and column during the P300 speller training will be adapted according to an iterative learning control law that was developed by the principal investigator of this study. The control law uses the previous number of flashes, as well as the participants' previous performance.
Control Group 1	Task difficulty adaptation according to Arvaneh et al. (2019)	In this group, the number of flashes per row and column during the P300 speller training will be adapted based on the participants' previous performance according to the approach used by Arvaneh et al. (2019).

Primary Outcome Measures

Name	Time	Method
Change in response times in the random dot motion task	Measured immediately before and after the training block in the experimental session (~1 hour and 30 minutes total)	The change in mean response times for correct responses in the random dot motion task from pre- to post-training is analysed.
Change in alpha power of EEG signals	Measured over the course of the training block in the experimental session (~1 hour and 30 minutes total)	The change in mean power of the alpha rhythm in the first 150ms after a non-target stimulus in the P300 speller task is analysed.
Change in accuracy in the random dot motion task	Measured immediately before and after the training block in the experimental session (~1 hour and 30 minutes total)	The change in mean accuracy, i.e. number of correct responses over total number of targets, in the random dot motion task from pre- to post-training is analysed.
Change in total power of EEG signals	Measured over the course of the training block in the experimental session (~1 hour and 30 minutes total)	The change in mean total power of EEG signals between 150ms to 550ms post-stimulus for target and non-target trials in the P300 speller task is analysed.
Length of training	Measured over the course of the training block in the experimental session (~1 hour and 30 minutes total)	The length of the P300 training phase is compared between groups.

Secondary Outcome Measures

Name	Time	Method
Response to NASA Task Load Index questionnaire	Measured immediately after the training block in the experimental session (~1 hour and 30 minutes total)	The response to the NASA Task Load Index questionnaire post-training is analysed. Participants evaluate mental, physical and temporal demand of the task, as well as their performance, effort and frustration regarding the task on a 21-point scale. A higher score means a higher perceived workload of the training.
Change in spelling accuracy in P300 speller task	Measured immediately before and after the training block in the experimental session (~1 hour and 30 minutes total)	The change in mean spelling accuracy, i.e. correctly identified letters over all letters in a run, from pre-training to post-training P300 speller runs is analysed.
Change in responses to questionnaire	Measured immediately before and after the training block in the experimental session (~1 hour and 30 minutes total)	The change in responses to the questionnaire about boredom and fatigue pre- and post-training is analysed. Participants answer the following 4 questions on a 10 point likert scale:(1) How tired are you now?, (2) How alert do you feel?, (3) How bored do you feel?, (4) Do your eyes feel tired? The higher the change in score of the responses, the more tiresome and/or boring the training session was for the participant.

Trial Locations

Locations (1)

Maynooth University; 🇮🇪 Maynooth, Co. Kildare, Ireland