Class Project
Experiment Design, Research, Data Collection, Data Analysis
November 2019 - December 2019
Urbana, IL
Humans are living in a lot of stimulus and spend tons of time sorting out and searching things. These processes require the quick adjustment of attention from one item to another.
There had been standard theories that visual search relies on memories, the accumulated information about items that the observer perceived before.
Based on the theories, it was assumed that search efficiency will be decreased if thevisual stimulus keeps changing over the search task.
Horowitz and Wolfe conducted an experiment that tested subjects’ ability to search a target from the visual scene, which was continually changing.
The results showed that searching efficiency is not decreased even though the visual scene was consistently changing, Horowitz and Wolfe proved that visual search does not rely on the memories.
They conducted a research that examined the effect of different distraction tasks on drivers’ performance by employing the EEG.
This paper suggested that any cognitive activity such as solving a math problem and daydreaming could impair driving performance, which requires a lot of visual and cognitive tasks.
The researchers paired the driving tasks, such as going straight and making left turn, with the audio distraction.
The study showed that the posterior brain, which is significant for visual attention and processing brain resources, was sacrificed when the drivers were distracted.
In this study, we aimed to explore the question of how humans perform visual tasks under different kinds of distractions.
Does your ability to find the target depend on the existence of distraction?
Does your ability to find the target depend on the type of distraction? (for our purposes, no distraction vs music vs podcast)
The experiment was conducted using 10 white paper clips, 9 of them were bent to different shapes and 1 of them was in the original form.
iphone XS electrical stopwatch was used to measure subjects’ response time.
Experimenters randomized the set-size and the target number of each trial, which were shown on the experimenter sheet.
Headphones were used during condition 2 and 3. For condition 2, Lo-Fi music was played. The list of the songs was as followed: Ocean Kisses by Galimatias, Take Care by Julian Avila, Day Break by Electric Mantis, Sleep In-Pryces Remix by Ukiyo, and We can’t Be Friends by Dream Koala.
For condition 3, a podcast, TechMeme Ride Home 31 October 2019 episode, was played.
The experiment was conducted for 3 conditions: with no distraction, with Lo-Fi music, and with podcast. One condition was consisted of 20 trials, resulting in a total of 60 trials.
Each condition was conducted in the order of condition 1, condition 2 and condition 3. There was no break between each condition.
The target number and the set size for each trial were randomly assigned by experimenter. The experiment was conducted in the right corner booth of the room 26 of Psychology Building at University of Illinois at Urbana-Champaign.
The experimenter chose 2 of 9 bent paper clips and named each clip as target number 1 and target number 2. The original paper clip was named as target number 3.
Once the subjects finished studying the targets, the experimenter tested them to see whether the subjects could identify the target correctly.
During the test, the subjects were required to call out the name of each target from two targets that the experimenter presented to them.
On each trial of condition 1, the subjects were required to close their eyes while the experimenter was placing the assigned numbers of distraction paper clips and the target paper clips. The set size and the position of the paper clips was randomized by the experimenter each trial.
Once the arrangement of the paper clips was done, the experimenter started the stopwatch and told the subject to open his or her eyes at the same time. The subject was required to call out the name of the target from a group of paper clips presented to him or her. The target response and RT(Reaction Time) of each trial were recorded by the experimenter.
For condition 2, the subject was required to repeat the same procedure of condition 1, while he or she was listening to the Lo-Fi music playlist, selected by the experimenter, with the headphones. Since the subject could not hear the cue sign to open his or her eyes, the experimenter tab the right hand of the subject to let the subject know he or she could start the trial.
Condition 3 followed the same procedure of condition 2 but used the podcast of TechMeme Ride Home October 31st 2019 episode instead of Lo-Fi music playlist. After the subject finished the whole experiment, he or she was asked to rate the distraction level of each condition, which was scaled from 1 to 10. The distraction rating of each subject was recorded by the experimenter.
Response Time, Response accuracy (which paperclip they choose vs what they were supposed to choose)
Simple Linear Regression, Multiple Regression
Predictions: what are the possible outcomes?
1. There is a positive effect.
The distraction is good to have or a search-oriented task.
2. There is a negative effect.
The distraction is not good to have for a search-oriented task.
3. There is no affect.
The distraction does not impact your performance on search-oriented tasks.
In this experiment, the target response, response time (RT), and distraction condition of subjects were recorded.
We conducted a simple linear regression to test the effect of different distraction stimuli, and a multiple regression to examine the effect of target type, set size, and condition number on the average RT.
A simple linear regression was conducted.Condition Number was a significant predictor of Error, b=2.115, t(37)=5.984, p=0.000.This result indicated that our distraction manipulation did affect our single regression.The mean of distraction rating of condition 3 with the podcast episode was the highest among three conditions. Condition 1 with no distraction stimuli showed the lowest mean of distraction rating.However, since we collected this result using a simple linear regression, we cannot conclude that one condition was more distraction than another condition.
A multiple regression was conducted.Target Type was a significant predictor of Error, b=0.247, t(219)=3.169, p=0.002. It showed that there was a significant difference in the RT for an original paper clip, named as 0, and RT for bent targets, named as 1.
A multiple regression was conducted.Set Size was a significant predictor of Error, b=0.104, t(217)=7.447, p=0.000.It showed that there was a significant difference in the RT for each target type.
A multiple regression was conducted.Condition Number was not a significant predictor of Error, b=-0.113, t(219)=-2.308, p=0.094.Different distraction stimuli did not have influence on the average response time (RT).
The condition 1, which was intended to be no distraction, actually ended up involving some noise from outside.
We, the experiments, also participated in our own experiment.
In terms of familiarity, the subjects might have got used to the experiment process and the targets, which could affect the average response time (RT).
The preference of each subject on the distraction might have affected the distracting level rating.
Even though there were some potential biases that we would like to address, these results suggest that audio distraction affects human ability of visual search but the types of distraction stimuli did not have significant influence on performance of visual search.
Although, a simple linear regression of different distraction rating showed that there was a significant difference in a way that subjects rated the distraction level for each condition.
It suggested that subjects felt more distracted during performing visual task in one condition than in another; however, it did not necessarily mean that one of the distraction stimuli was more distracting due to the limitation of the data analysis methodology.
