Toward a Theoretical Framework for Predicting Overall Quality Effects of Interruptions on Content Production Tasks

Nicole Werner

Advisor: Deborah A. Boehm-Davis, PhD, Department of Psychology

Committee Members: Carryl Baldwin, Robert Youmans

Research Hall, #92
October 23, 2014, 11:00 AM to 08:00 AM

Abstract:

This dissertation explores the affect of interruptions on the overall quality of work and proposes a new framework for predicting interrupted task performance. Five experiments are described that set about to determine if there is an effect of interruptions on content generation, and to explore a potential theoretical explanation for the underlying cognitive mechanisms that are driving this behavior.

The following chapters present two sets of studies that sought to 1) determine whether interruptions affect the overall quality of work and 2) explore what mechanisms underlie interrupted task performance related to overall quality. We used a new interruption experimental paradigm in order to create a task where overall quality could be defined beyond the number of errors made, or the length of time to complete the task. Essay writing and essay outlining provided the necessary paradigm as a complex, creative thought task.

The first set of studies focused on whether interruptions affected the overall quality of work using an essay writing and outlining paradigm. The results suggested that interruptions during a writing task negatively impacted the overall quality of the writing. Moreover, when interruptions were experienced only during the planning of the essay, the overall quality of the essay suffered. Counter to findings from interrupted task performance during procedural tasks, the second study determined this effect persisted even if participants were given as much time and they wanted to complete each outline and essay. It was not clear whether current theoretical models of interrupted task performance could account for the resulting performance.

The second set of studies sought to expand on this work by investigating underlying cognitive mechanisms might be driving the detrimental effect on overall quality found in the first paper. In a series of three studies, we first quantified and compared outlines to determine how interruptions affect the outlining process and to explore the relationship between an outline created when interrupted and the resulting essay. In the first experiment, we found that interruptions led to less fully developed outlines and that this carried over to the essay. Next, we wanted to further explore the cognitive mechanisms underlying this performance through a Verbal Protocol Analysis of the outlining task during interruption. We were specifically interested in how people disengaged from the outlining task to start the interrupting task, and then the process used to resume and reengage the primary task when the interrupting task ended. We found support for a combination of underlying processes including a memory for the task location and the activation associated with the goal memory. We also found that unlike resumption to a procedural task, participants had difficulty reconstructing pre-interruption context at the point of resumption. This was likely due to the task context being largely internal, rather than driven by the task itself. These results suggested a spreading activation theoretical account for interrupted task performance based on the activation and associative memory models. Last, we provided further support for spreading activation theory as a model for interrupted task performance in a content creation task by making predictions for potential mitigation based on this theory. We compared different interruption lengths with short or long interruptions. We found that frequent interruptions lead to more disruption in a content production task compared to one long interruption or no interruptions. It is possible that a longer interruption led to complete decay of activation associated with the primary task such that participants were forced to start from the beginning, recreating the initial context through spreading activation from the original node. This is contrary to the frequent interruption conditions in which participants tend to resume to the point at which they were interrupted, leading to a change in internal context when activation is spread from the resumption point as the new source node.