Estimate the Cognitive Load Using Electrocardiographic Measure: A Human-AI Collaborative Task

We study how people's cognitive load changes when working with AI in a delegation setting. To begin, introduce the experimental procedure to the participant, and ensure they're fully informed about the study. Prepare the electrocardiogram data acquisition system and gather all necessary materials, including the pre-gelled, disposable electrocardiogram electrodes.

Connect the data acquisition system to the electrocardiogram data receiver module. Using the male to female ribbon cable, establish a connection between the data acquisition system and the computer used for stimulus presentation. Using an ethernet cable, establish a connection between the data acquisition system and the computer used for data recording.

Plug the system into a stable power supply. Now, clean the participant's skin with alcohol wipes at three sites. Below the left rib, below the right clavicle, and on the right abdomen.

Attach one disposable electrocardiogram electrode patch to each clean site. Then attach the lead wire connector securely to the electrocardiogram transmitter, and connect the metal clip of the lead wire to the corresponding electrode. Secure the electrocardiogram transmitter to the participant's body using stretchy hook-and-loop fastener straps and adjust the strap tension to ensure the transmitter stays in place and remains comfortable for the participant.

Turn on the switch of the electrocardiogram transmitter and power on the electrocardiogram data acquisition system. Check the wireless connection status between the transmitter and the receiver module of the data acquisition system. Now, launch the electrocardiogram data acquisition software.

Create a new graph file. Select the electrocardiogram receiver module from the device list. Choose the appropriate electrocardiogram recording channel and click okay in the newly displayed transmitter connection status window.

Click start in the data acquisition software interface to begin electrocardiogram data collection. Verify that the electrocardiogram waveform appears stable on the screen. Now, instruct the participant to begin the experimental task.

After the task is completed, click stop in the data acquisition software to end the recording session. Detach the lead wire clips from the electrode patches. Loosen the stretchy hook-and-loop fastener straps and remove the electrocardiogram transmitter from the participant's body.

Instruct and assist the participant in gently removing the disposable electrode patches by hand. Next, open the recorded electrocardiogram data file in the data analysis software. Select the electrocardiogram channel and apply digital filtering.

In the menu bar, navigate to transform followed by digital filters and comb band stop. Then in the dialog box, select fixed at under base frequency. Enter 50 hertz and click okay to apply the filter.

Next, click transform in the menu bar. Select digital filters, then FIR, followed by band pass. In the dialogue box, under low-frequency cutoff, select fixed at and enter one hertz.

Under high-frequency cutoff, select fixed at and enter 35 hertz. Now, create focus areas based on the experimental design, Generate and export heart rate data by selecting analysis, followed by hemodynamics and ECG interval extraction from the menu bar. In the analyze option, choose focus areas only and in the display results as option, select Excel spreadsheet only.

Click okay to confirm and complete the export. To generate heart rate variability data, open analysis and select HRV and RSA followed by multi-epic HRV statistical. In the ECG channel field, select the pre-processed electrocardiogram channel and under extract HRV statistics for choose focus areas.

Go to output results to, select spreadsheet, and click okay to generate and export the time domain results. Finally, open analysis again and select HRV and RSA followed by multi-epic HRV and RSA spectral. In the ECG channel field, select the pre-processed electrocardiogram channel.

Under extract HRV and RSA for, choose focus areas. In output type, select Excel spreadsheet only, and click okay to export the frequency domain results. Mean heart rate was significantly lower under the difficult image condition compared to the easy image condition.

High-frequency heart rate variability was significantly higher during the difficult image condition compared to the easy image condition. Root mean square of successive differences, followed a similar trend and was significantly higher during the difficult image condition. Participants delegated significantly more frequently under the difficult image condition than the easy image condition.

We found that people adjust their cognitive load by delegating tasks when working with AI.In future, we will look at how different co-learning approaches affect outcomes when human and AI learn together.