EEG-based Evaluation of Cognitive Workload Induced by Acoustic Parameters for Data Sonification

Abstract

Data Visualization has been receiving growing attention recently, with ubiquitous smart devices designed to render information in a variety of ways. However, while evaluations of visual tools for their interpretability and intuitiveness have been commonplace, not much research has been devoted to other forms of data rendering, e.g., sonification. This work is the first to automatically estimate the cognitive load induced by different acoustic parameters considered for sonification in prior studies [9,10]. We examine cognitive load via (a) perceptual data-sound mapping accuracies of users for the different acoustic parameters, (b) cognitive workload impressions explicitly reported by users, and (c) their implicit EEG responses compiled during the mapping task. Our main findings are that (i) low cognitive load-inducing (i.e., more intuitive) acoustic parameters correspond to higher mapping accuracies, (ii) EEG spectral power analysis reveals higher band power for low cognitive load parameters, implying a congruent relationship between explicit and implicit user responses, and (iii) Cognitive load classification with EEG features achieves a peak F1-score of 0.64, confirming that reliable workload estimation is achievable with user EEG data compiled using wearable sensors.

Reliability of Mental Workload Index Assessed by EEG with Different Electrode Configurations and Signal Pre-Processing Pipelines

“…Various pre-processing methods are currently proposed in the literature to extract MWL indicators from EEG signals, and a consensus is still missing among researchers. The bandpass filtering is typically used in most papers but with different cut-off frequencies [ 20 , 23 , 24 ]; the major artifacts are typically removed with Independent Component Analysis (ICA) [ 25 , 26 ], Artifact Subspace Reconstruction (ASR) algorithms [ 20 , 27 ] or other methods [ 28 ]; the signal is mainly re-referenced to the average of the electrodes [ 23 , 27 ] or the average of the mastoid electrodes [ 20 ]; the channel rejections are performed automatically [ 29 ] or manually [ 30 ]. Although some more general pipelines for EEG signal analysis exist, they are quite broad and not universally adopted [ 31 ].…”

Comparing Performance and Preference of Visually Impaired Individuals in Object Localization: Tactile, Verbal, and Sonification Cueing Modalities

“…Contrarily to the positive effectiveness of verbal cueing, sonification poses a few challenges, mainly due to the lack of crucial information about the direction. Also, the continuous beeping might be annoying, and the cognitive workload required to convert the beeping sound into spatial information might contribute to the observed delay, as suggested by [20]. While target sonification might prove advantageous for sighted individuals, particularly in scenarios involving intricate visual guidance such as surgery [21], for the visually impaired population, sonification may present challenges due to the inherent need for comprehensive auditory cues and efficient cognitive processing.…”

Gender and emotion recognition with implicit user signals

“…Many works perform ER with implicit behavioral signals such as eye movements, EEG, Electromyogram (EMG), Galvanic Skin Response (GSR) [27], [28], [29], [30], [31] etc. However, very few works estimate soft biometrics such as gender, cognitive load and personality traits with such signals [32], [33], [34]. Also, while some works isolate emotion and gender differences in eye movement and EEG responses to emotional faces [24], [25], [35], [36], these differential features are never utilized for gender prediction.…”