realizados por un remitente: Alerta de Envío
9.1 Acerca de los reportes
9.2.1 La pantalla Selección de reporte
9.2.1.2 Para crear un nuevo reporte con un diseño predefinido
Confirmation bias is a prevalent cognitive bias in decision making. This bias results in errors in decision making by inappropriately bolstering a believed hypothesis. Confirmation bias can also affect visual search by restricting the searcher’s attention to one visual hypothesis. Mitigating a confirmation bias during a visual search is possible but it is unknown how such a mitigation would affect the searcher’s search patterns. Finally, current research indicates that confirmation bias can be detected through the use of EEG data.
III. Methodology
3.1 Chapter Overview
This chapter describes the outline of a human-participant visual search experiment and the process used to analyze the recorded data. First, the chapter discusses the research questions and hypotheses. Next, a description of the experiment which includes participant demographics as well as details about the various factors and variables present in the experiment is presented. This is followed by a description of how the results will be analyzed. This section will include details about the statistical tests to be performed as well as the machine learning approach used. Finally, a summary of this chapter is provided.
3.2 Background
Humans often default to a confirmatory method of searching because it generally requires less mental processing, it is a simple way to search for a single target, and it is often the most efficient way to search [10], [65], [73]. Because it is simple to perform, humans will perform a confirmatory search even when a confirmatory method of searching is not the most efficient way to search [74]. However, because it is still possible to perform a confirmatory search while also being efficient and because it is possible to perform a non-confirmatory search while being inefficient, this research focuses on encouraging efficient searches rather than discouraging confirmatory searches.
3.2.1 Previous Work
The experiment conducted to answer these research questions was an extension of Rajsic, Wilson, and Pratt’s 2015 and 2017 experiments [73], [74]. As an overview, Rajsic’s 2015 experiment was a visual search task in which participants were presented
with eight letters arranged in a ring. The letters on the screen could be one of two colors. Participants were instructed to search for a specific letter, called the target letter, and were given an example of what color the letter could be, or the template color. The target letter would appear only once amongst the other letters. The participants were to press a certain key if the target letter’s color matched the template color and to press a separate key if the target letter’s color did not match the template color. An example of Rajsic’s 2015 experiment can be seen in Figure 8
Figure 8. Rajsic’s 2015 experiment. The instructions presented to the subject, along with a template color match and mismatch, and the predicted results can be seen.
The results of Rajsic’s 2015 experiment showed that participants performed a confirmatory search by searching all of the template-matching colored letters first, even when there were more template-matching colored letters than template-mismatching colored letters.
to perform a non-confirmatory, or an efficient, search. This cost came in the form of increasing the time necessary to search for the target letter. The experiment was identical to the 2015 experiment except that instead of letters being present on the screen, there were now eight colored circles. To make a letter appear the participant would have to visually fixate on the circle. Under these search conditions the participants were able to prioritize their search towards the color with the smallest amount of circles present, thus reducing the confirmation bias present in the visual search. An example of Rajsic’s 2017 experiment can be seen in Figure 9.
Figure 9. Rajsic’s 2017 experiment.
3.3 Research Questions
The objective of this research is to determine whether an inefficient search during a visual search can be detected and subsequently mitigated. To complete this objective, the following research questions are investigated.
3.3.1 Research Question 1 - Categorizing Natural Behavior
What visual search patterns do participants naturally use during a visual search task?
Hypothesis: The majority (> 50%) of participants will naturally resort to an inefficient Visual Search Pattern (VSP).
3.3.2 Research Question 2 - Behavior Detection
Can physiological signals such as Electroencephalography (EEG), Electrooculography (EOG), and Electrocardiography (ECG) be associated with an efficient visual search?
Hypothesis: Physiological signals can differentiate a participant performing an efficient visual search from a participant performing an inefficient visual search. Research Objective: Develop a machine learning model that receives physiological
data and is able to determine an efficient visual search with an equal-class-weighted classification accuracy of greater than 50%.
3.3.3 Research Question 3 - Behavior Mitigation
For a participant who is performing an inefficient search, can mitigation techniques change the participant’s search patterns to an efficient search pattern that will persist for the remainder of the search tasks?
Hypothesis: By applying the mitigation techniques of a nudge, a hint, and by teaching the participant how to perform an efficient search, a participant will perform an efficient search pattern for the remainder of the search tasks.