Humor or Rationality? The Neural Mechanisms of How Agent Type and Language Style Influence Satisfaction with Ride-Hailing Service Failure Recovery

We explore neural mechanisms of agent types and language styles on ride-hailing recovery satisfaction. Traditional approaches lead direct neural evidence. This protocol captures real-time brain activity to review underlying cognitive processing mechanisms.

To begin, conduct a preliminary briefing after the participant arrives at the laboratory. Introduce the core equipment, materials, and the complete experimental procedure step-by-step. Also clearly explain the experimental tasks, the estimated duration, and important precautions.

Provide a written informed consent form to the participant. Instruct the participant to carefully read the form and sign it before the experiment commences. Instruct the participant to wash their hair with neutral shampoo in the designated area to remove scalp oil, dust, and impurities that may affect electrode conductivity.

Using the laboratory-provided hair dryer, thoroughly dry the hair to ensure no damp areas remain. Then guide the participant into the dedicated EEG recording laboratory maintained at a temperature of 22 degrees Celsius with dim lighting and equipped with electromagnetic shielding and sound insulation. Instruct the participant to sit in the comfortable chair about one meter from the center of the screen.

Using 75%medical alcohol cotton balls and non-irritating facial cleanser, gently wipe the specified facial areas of the participant. Position the electrode cap equipped with 64 silver silver chloride electrodes on the participant's scalp according to the international 10-20 system. Align the midline of the cap with the nasion to inion line.

Then place the CZ electrode at the vertex. Verify that FP1 and FP2 are approximately two centimeters above the nasion and that T3 and T4 are approximately two centimeters above the ear tragus. Fill all electrode wells with conductive gel.

Then secure external electrodes with medical tape. Place the reference electrode on the tip of the nose. Position the vertical electrooculogram electrodes one centimeter above and below the left orbital socket.

Then place the horizontal electrooculogram electrodes one centimeter lateral to the outer canthus of each eye and the M1 and M2 electrodes on the left and right mastoids, respectively. Now, adjust the chin strap of the electrode cap to a snug, yet comfortable fit. Verify that the cap is not twisted or rotated and confirm its final position.

Switch the recording software to impedance check mode. Using a blunt tip syringe, fill the electrode wells with conductive gel. Gently part the hair at each site to ensure full scalp contact.

Adjust the electrodes until each channel impedance drops below 10 kilo ohms. Verify that acceptable values are indicated by blue or dark color on the monitor. Now, inform the participant that the experiment will take place in a closed and quiet environment.

Instruct them to remain relaxed, maintain a stable sitting posture, minimize body movements, and stay focused throughout the session. Then present the stimuli on a 22 inch liquid crystal display monitor. Set the screen refresh rate to 60 hertz.

Insert the E-Prime hardware key into a universal serial bus port on the computer. Double click the E-Studio icon to launch the software. Then select file and choose open to load the project file.

On the E-Run startup interface, enter the subject number, the session number, and other required information. Now click on Tools and E-Run. Then choose run to execute the practice session.

Ensure the participant understands the instructions and runs the 10 practice trials identical to the main experiment. After completion, ask whether the participant clearly read and understood the stimuli. Proceed to the instruction screen after the practice trials once the participant confirms they are ready.

Instruct the participant to press the space bar to begin the main experiment, consisting of 120 trials divided into two blocks of 60 trials each with a mandatory two minute rest between blocks. Monitor the session progress from an adjacent room using a monitor screen. Upon completion, ensure the software automatically saves the data file.

Now, remove the electrode cap from the participant. Instruct the participant to clean residual conductive gel from their hair and skin and ensure no residue remains. Confirm that the participant experiences no discomfort after cleaning.

Conclude the experiment and provide the remuneration. Visually inspect the EEG data and exclude epochs with significant baseline drift or severe noise artifacts. Set the vertical electrooculogram channel as the reference.

Apply the ocular artifact reduction algorithm in the scan software to correct blink and eye movement artifacts. Next, using event markers inserted by the E-Prime program, segment the continuous data into epochs timelocked to stimulus onset. Define the time window from minus 200 milliseconds to plus 800 milliseconds.

Define the 200 millisecond pre-stimulus interval as baseline and apply baseline correction in the scan software. Reject epochs where the signal amplitude exceeds plus or minus 100 microvolts within the analysis window to eliminate artifacts. Compute the average signal from the left and right mastoid electrodes.

Re-reference all channels offline to this mastoid average. For each participant, average all artifact-free epochs within the same stimulus condition to obtain condition-specific event-related potential waveforms. Then apply a 30 hertz low pass filter with 24 decibels per octave rolloff to the averaged waveforms.

Combine individual averages across all participants to create grand average event-related potential for each condition. Satisfaction ratings showed a significant main effect of service agent with satisfaction for the human agent being higher than for the AI agent. A significant main effect of language style was also found, with satisfaction in the humorous condition higher than in the rational condition.

A significant interaction between service agent and language style was observed. In the humorous language condition, satisfaction was higher for the human agent than for the AI agent. The N2 component showed a significant main effect of language style with the rational language style eliciting a more negative N2 than the humorous language style.

A significant interaction between service agent and language style was observed for the P300 component. In the humorous language condition, the human agent elicited a more positive P300 than the AI agent. In the AI agent condition, the rational language style elicited a more positive P300 than the humorous language style.

The protocol measures users'neural and behavioral response to agent types and language styles in ride-hailing service recovery strategies. Occurred stimulus timing, precise electrode placement, low impedance, and rigorous artifact rejection are critical for obtaining reliable EIP data. Future research could explore cross-cultural effects, personality influence, and work with us in AI human recovery communication.