The Problem State: A Cognitive Bottleneck in Multitasking

Ganesh Pitchiah - Y9213

Proposal

Motivation

Cognitive Architecures are used to reproduce the behaviour of the modelled system (human), in a way that timely behavior (reaction times) of the architecture and human subjects can be compared in detail. This helps gain points in favor of a proposed cognitive model.
This particular paper [Borst et al. '10] too designs an expirement to compare reaction times of an already implemented theory of Threaded Cognition [Anderson '07] in ACT-R and human participants. However, the focus being to identify the "Problem State" resource as a bottleneck in multitasking apart from the existing ones.

Review of Literature

Multitasking is the ability to handle more than one task at a time by an individual. Psychologists have been formally investigating multitasking behaviour at least since the 1930's. Various cognitive models of multitasking have been developed to predict when and how task interfere. To unify several areas of multitasking, a new theory of multitasking behavior - Threaded Cognition [Salvucci & Taatgen '08] was proposed.

The key assumption related to multitasking in threaded cognition is: several tasks can be active at the same time, but a particular resource can only be used by a single task at the same time. Hence, it predicts interference when 2 or more task concurrently need a particular resource; the resource will act as a bottleneck and delay the execution of the combined process.

Problem state [Borst et al. '07] resource is used for storing intermediate information that is necessary for performing a task. For instance, while solving 2x - 5 = 8, problem state can be used to store the intermediate result 2x = 13. It can be compared to a buffer with limited capacity. Though, [Salvucci & Taatgen '08] discusses 4 bottlenecks, the present authors identify problem state resource as a different and an important source of interference. And, the experiments are designed to show problem state resource indeed acts as a bottleneck in multitasking.

ACT-R (Adaptive Control of Thought—Rational) is a cognitive architecture which aims to define the basic and irreducible cognitive and perceptual operations that enable the human mind. In theory, each task that humans can perform should consist of a series of these discrete operations. Threaded cognition has already been implemented in ACT-R [Anderson '07].

Keywords: multitasking, interference, problem state, threaded cognition, ACT-R

Methodology

  1. In this experiment, a subtraction task and a text entry task have to be performed concurrently by alternating between the two.
  2. Both tasks are presented in 2 verions: one that requires maintaining a problem state and one that doesn't.Generating, four possible conditions in any given trial.
  3. The experiment consists of three blocks. Each block has 4 sets of three trials per condition, i.e, 36 trials per subject for 15 such.
  4. The reaction times of every step in each task is measured.
  5. The accuracy and response times of the experiment and model will be compared to determine the interaction effects.
    If, indeed the problem state resource is the bottleneck, then the results should be comparable in all four conditions.

Resources & References

ACT-R and Models built on ACT-R used in the experiment.
Borst, Jelmer P., Niels A. Taatgen, and Hedderik van Rijn. "The problem state: A cognitive bottleneck in multitasking." Journal of Experimental Psychology: Learning, memory, and cognition 36.2 (2010): 363.
Borst, J. P., and N. A. Taatgen. "The costs of multitasking in threaded cognition." Proceedings of the Eighth International Conference on Cognitive Modeling. 2007.
Salvucci, Dario D., and Niels A. Taatgen. "Threaded cognition: an integrated theory of concurrent multitasking." Psychological Review 115.1 (2008): 101.
Anderson, John R. How can the human mind occur in the physical universe?. Vol. 3. Oxford University Press, USA, 2007.