A computational theory of intention selection

TY - ABST

T1 - A computational theory of intention selection

AU - Oren, Franziska

AU - Kyllingsbæk, Søren

AU - Grünbaum, Thor

PY - 2018/1

Y1 - 2018/1

N2 - In a buzzing, vibrant world, we form a variety of different intentions in the course of a day. Thus, at any one time, a large number of intentions is represented in memory and many of these may be relevant in the current context. In order to function well as a human being, successful control of action by decision-making requires a solution to the computational problem of how the right intention is selected, out of a multitude of intentions in memory, for execution at the right time. Without such a selection mechanism, we could not simultaneously form and store several intentions for future actions, we would lose our ability for long-term planning, and our psychological and practical life would lose its structure and stability. 
Previously proposed theories of the formation of intentions in the psychological and philosophical literature (e.g., Pacherie, 2008), as well as established paradigms in the prospective memory literature (e.g., Einstein & McDaniel, 1990), only account for how a single intention, or a single prospective memory task, is produced and implemented in the current context. In contrast, we propose a Computational Theory of Intentional Action Selection (CTIAS) where the focus is on the competition among several relevant intentions for selection. CTIAS is a race model that explains in mathematical terms how standing distal intentions are selected from long-term memory and transformed into occurrent distal intentions in working memory. The underlying idea is that all components of all competing standing intentions are racing against each other, with a winner at the end that is selected for execution. Components from other standing intentions that finish the race later are lost. In this way, CTIAS effectually decides which intention is occurrent in working memory, and therefore, upon which intention a subject is most likely to act upon at any given moment in time.

AB - In a buzzing, vibrant world, we form a variety of different intentions in the course of a day. Thus, at any one time, a large number of intentions is represented in memory and many of these may be relevant in the current context. In order to function well as a human being, successful control of action by decision-making requires a solution to the computational problem of how the right intention is selected, out of a multitude of intentions in memory, for execution at the right time. Without such a selection mechanism, we could not simultaneously form and store several intentions for future actions, we would lose our ability for long-term planning, and our psychological and practical life would lose its structure and stability. 
Previously proposed theories of the formation of intentions in the psychological and philosophical literature (e.g., Pacherie, 2008), as well as established paradigms in the prospective memory literature (e.g., Einstein & McDaniel, 1990), only account for how a single intention, or a single prospective memory task, is produced and implemented in the current context. In contrast, we propose a Computational Theory of Intentional Action Selection (CTIAS) where the focus is on the competition among several relevant intentions for selection. CTIAS is a race model that explains in mathematical terms how standing distal intentions are selected from long-term memory and transformed into occurrent distal intentions in working memory. The underlying idea is that all components of all competing standing intentions are racing against each other, with a winner at the end that is selected for execution. Components from other standing intentions that finish the race later are lost. In this way, CTIAS effectually decides which intention is occurrent in working memory, and therefore, upon which intention a subject is most likely to act upon at any given moment in time.

M3 - Conference abstract for conference

T2 - 5th International Conference of Prospective Memory

Y2 - 3 January 2018 through 6 January 2018

ER -