Watching TV conceptualized as a decision-making task

On average, Americans spend about 3 hours a day watching stories on screens—about 20% of their waking life. This time is not entirely passive—it’s filled with often complex cognitive tasks, including processing and interpreting stories, choosing what to watch, whether to watch and how, and additionally how to coordinate consuming content with other tasks, like eating, cleaning, socializing, and sleeping.  All in all, watching television is a busy mental time.  I will survey a few of the clearest examples of cognitive tasks that have been important to me over my seven years as a data and machine learning scientist at Netflix. I will then focus in on one important decision: when to keep watching a show you may or may not enjoy.   When people watch movies in the theater, they are largely committed: the movie starts, and they watch it until it is completed.  However, when people watch at home, they may or may not complete any given title. We can consider the act of continuing the act of watching a title as an extended decision-making task, in which the title plays the role of a time-varying stimulus. I will take the audience through the impact of different ways of construing task parameters such as time (e.g., wall time, point in the content, or content seconds viewed), viewer decisions (e.g., a continuous alternating decision, or a single go/no-go decision), and individual differences, as well as modeling frameworks (say, drift diffusion, or utility models). Finally, I’ll loop back around to comment briefly on how cognitive science can be informed by contact with real-world problem spaces like those we face in industry settings.

Sponsorship of an event does not constitute institutional endorsement of external speakers or views presented.