Keywords

1 Introduction

By realizing the Computerized Brain Interfaces for Adaptive Learning and Assessment project, the final intention of the authors is to make progress in the process of continuous improvement of the quality of education, which identifies the public universities of Spain, based on the premises of collaboration between teachers and students. Therefore, two fields are considered and related as it illustrates the Fig. 1.

Fig. 1.
figure 1

Schematic workflow of a BCI application - an online adaptive learning environment based on EEG data

Full size image
  1. 1.

    Adaptive learning and assessment, which is the focus of the research project, is based on the search of methods to adjust the level of demand (WML, fatigue, stress…) to which the student is subjected to the difficulty of tasks, tests or lessons taught. An illustrative scheme is presented in the Fig. 2.

    Fig. 2.
    figure 2

    Illustrative scheme of an adaptive learning method and its relation with EEG

    Full size image
  2. 2.

    The electroencephalogram is a noninvasive method to measure the brain wave pattern for the identification of electrical activities in the brain. The distribution of the Alpha, Beta and Gamma brain waves will provide abundant information on the student’s brain state while exposed to differentiated learning strategies. This way, researchers can determine how brain patterns reflect the student’s level of learning and when the brain patterns confirm that the student has mastered the skills to be acquired.

2 Context and Purpose

Context.

Until now, learning environments primarily attend to a student’s performance and behavior. These systems [1, 2] aim to diagnose the development of the amount of knowledge acquired by a student based on response to questions, selection of links or errors during problem solving. However, online predictions of workload, fatigue and attention states may be used to tailor the learning material presented to the subject [3, 4] in a specific and personalized way, helping to maintain the burden of work coped by each student within an optimal range for learning.

The current neurological theory and the more recent cognitive studies of learning allow to set a relation between the brain activity (the brain waves and the areas of the brain where they occur), the difficulty of the learning tasks and the level of performance attained in the activities. The method proposed in this project is based on the monitoring of the brain patterns of the students, through EEG, when they perform tasks of different complexity, to reflect the level of learning and mastery of the skills to be acquired.

Two critical factors have propelled the realization of this project. The first is the current situation of the technological market, more specifically in relation to leisure and video games, which offers the possibility of acquiring simple non-intrusive low-cost EEG measurement devices. The second, the previous execution of projects within the ETSIAE related to the development of cognitive workload models in environments similar to those experienced by an air traffic controller [5], as well as the use of electroencephalography in practical applications such as drone control [6].

Purpose.

The main objective is to analyze the feasibility of applying low-cost electroencephalogram (EEG) based computer interfaces to adaptive learning and assessment strategies and systems. In conclusion, to evaluate the feasibility of using brain waves to improve adaptive learning tools. Besides, the project has two sub-objectives:

  1. 1.

    Evaluate the usability of low-cost EEG devices for the measurement of brain waves for educational and research purposes.

  2. 2.

    Develop neurological signal-processing algorithms that allow the assessment of the student’s cognitive state in the face of learning tasks of different complexity.

3 Methodology and Results

Methodology.

A thorough analysis of previous studies has allowed the identification of two promising research lines to incorporate EEG technology in adaptive learning, with the project activities focusing so far on the first. One is the determination of working memory load (amount of mental resources used to execute a specific task), the other the determination of the emotional state and its influence on learning [7,8,9].

The workload during learning is the result of an interaction between the complexity of the contents to learn, the instructional design and the previous knowledge of an apprentice [10]. So far, methods for detecting workload are based primarily on subjective workload ratings [11], or on dual-task procedures [12]. Nonetheless, EEG measures can be used to estimate the amount of workload of each student during a learning session, based on the following hypothesis. Increased workload leads to increased activity in the theta frequency band (synchronization) in the electrodes of the frontal zone [ 13, 14 ] and to a decrease in activity of the alpha frequency band (desynchronization) on the parietal and occipital area [ 13, 15 ].

To verify the feasibility of measuring mental workload with low-cost EEG devices, a set of experiments have been carried out using accessible technologies and a learning environment based on on-line games [16] of incremental difficulty that do not require previous knowledge by the student. These mini-games test a set of basic cognitive abilities. For example, the game Shape Tracking, tests the ability to follow an element on the screen, with each level increasing in difficulty. For the study, level 1 was labeled as low workload, and the maximum level that each subject has reached was high workload.

Measurements were taken with the device Emotiv Insight [17], a wireless 5-channel EEG handset that records brain waves and translates them into meaningful data. It uses a patented polymer dry sensor that is safe to use and offers great electrical conductivity. Signal processing was performed with the SW EEGLAB [18], an interactive Matlab toolbox for processing EEG, MEG and other continuous event-related electrophysiological data. Figure 3 shows the measurement device and a snapshot of the experiments performed. Five subjects and three difficulty levels were tested. Figure 4 illustrates the results obtained when calculating the PSD (Power Spectral Density) in θ and α bands.

Fig. 3.
figure 3

Measuring equipment (left) and experiments performed (right)

Full size image
Fig. 4.
figure 4

Example of processed results (PSD of subject 1 in θ band for low workload. Test 1)

Full size image

Results.

Analyzing the behavior of the PSD theta frontal for the different tests and subjects, it is proved that PSD increases in high workload conditions, compared to data extracted for low workload. In addition, the PSD theta frontal variation in subjects is smaller as the session time increases. These data corroborate the hypothesis assumed.

The experiments realized satisfy the initial objectives of the project, since they have allowed the familiarization with the EEG measurement device, proving its value for the intended purpose; and have allowed to identify a methodology and indicators that allow to estimate the workload associated with learning tasks. It is therefore confirmed that this research line has the potential to be further developed in upcoming studies.