Limits...
Local dynamics in decision making: The evolution of preference within and across decisions.

O'Hora D, Dale R, Piiroinen PT, Connolly F - Sci Rep (2013)

Bottom Line: Within decisions, perceived alternatives compete until one is preferred.These decision spaces evolved through the experiments, as participants learned which options to choose.This visualisation approach provides a potential framework for the analysis of local dynamics in decision-making that could help mitigate both theoretical disputes and disparate empirical results.

View Article: PubMed Central - PubMed

Affiliation: School of Psychology, National University of Ireland, Galway, University Road, Galway, Ireland. denis.ohora@nuigalway.ie

ABSTRACT
Within decisions, perceived alternatives compete until one is preferred. Across decisions, the playing field on which these alternatives compete evolves to favor certain alternatives. Mouse cursor trajectories provide rich continuous information related to such cognitive processes during decision making. In three experiments, participants learned to choose symbols to earn points in a discrimination learning paradigm and the cursor trajectories of their responses were recorded. Decisions between two choices that earned equally high-point rewards exhibited far less competition than decisions between choices that earned equally low-point rewards. Using positional coordinates in the trajectories, it was possible to infer a potential field in which the choice locations occupied areas of minimal potential. These decision spaces evolved through the experiments, as participants learned which options to choose. This visualisation approach provides a potential framework for the analysis of local dynamics in decision-making that could help mitigate both theoretical disputes and disparate empirical results.

Show MeSH

Related in: MedlinePlus

Experimental task and learning data.The top panel provides a schematic of the experimental procedure. The lower left panel depicts mean probability of choosing a High option in a High/Low decision across participants on consecutive blocks of 6 decisions in each of the three experiments (error bars denote standard errors). The shaded area indicates probability below 80%, the threshold used to infer that participants learned to choose the High option, and the dashed line indicates 50%, the choice probability expected by chance. The lower right panel is a bubble plot of sensitivity to relative reward measured by the log2 probability of choosing a High option. As relative reward increased (the horizontal axis), more participants reliably chose the High choice in High/Low decisions. The size of each point is determined by the number of participants that obtained that value controlled for the number of participants in that experiment (i.e., probability density of that value within each experiment). The shaded area and black dashed line correspond to the same values as in the left panel. The blue dashed line indicates the mean log2 probability at each level of relative reward.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3713532&req=5

f1: Experimental task and learning data.The top panel provides a schematic of the experimental procedure. The lower left panel depicts mean probability of choosing a High option in a High/Low decision across participants on consecutive blocks of 6 decisions in each of the three experiments (error bars denote standard errors). The shaded area indicates probability below 80%, the threshold used to infer that participants learned to choose the High option, and the dashed line indicates 50%, the choice probability expected by chance. The lower right panel is a bubble plot of sensitivity to relative reward measured by the log2 probability of choosing a High option. As relative reward increased (the horizontal axis), more participants reliably chose the High choice in High/Low decisions. The size of each point is determined by the number of participants that obtained that value controlled for the number of participants in that experiment (i.e., probability density of that value within each experiment). The shaded area and black dashed line correspond to the same values as in the left panel. The blue dashed line indicates the mean log2 probability at each level of relative reward.

Mentions: Across three experiments, the ratio of the high-point reward to the low point reward was increased: Experiment 1 (7/5; n = 34), Experiment 2 (10/5; n = 37) and Experiment 3 (20/5; n = 55). Learning was faster and more reliable as the high-point value increased. On average, participants learned more quickly to choose the higher of the two options in the 20/5 experiment than in the other two experiments; after just 12 decisions, participants chose the high-value symbol on over 80 percent of High/Low decisions. By halfway through the experimental session (18 decisions), the high-value symbol was chosen reliably in High/Low decisions in all three experiments (see Fig. 1, bottom-left panel). The high-point value also affected the proportion of participants who learned to consistently choose the High options. The bottom-right panel of Fig. 1 provides the distribution of high-point/low-point choice ratio across participants in each experiment. The log2 of the ratio of the probability of choosing the high choice to the probability of choosing the low choice on any High/Low decision (i.e., log2(pHigh/pLow)) is employed as a measure of the degree to which the participant reliably chose High options in High/Low decisions (in the literature on basic learning principles, the log2 ratio is employed as an index of relative allocation of behavior to two independent responses and it has been shown to be sensitive to relative magnitude of reinforcement52.


Local dynamics in decision making: The evolution of preference within and across decisions.

O'Hora D, Dale R, Piiroinen PT, Connolly F - Sci Rep (2013)

Experimental task and learning data.The top panel provides a schematic of the experimental procedure. The lower left panel depicts mean probability of choosing a High option in a High/Low decision across participants on consecutive blocks of 6 decisions in each of the three experiments (error bars denote standard errors). The shaded area indicates probability below 80%, the threshold used to infer that participants learned to choose the High option, and the dashed line indicates 50%, the choice probability expected by chance. The lower right panel is a bubble plot of sensitivity to relative reward measured by the log2 probability of choosing a High option. As relative reward increased (the horizontal axis), more participants reliably chose the High choice in High/Low decisions. The size of each point is determined by the number of participants that obtained that value controlled for the number of participants in that experiment (i.e., probability density of that value within each experiment). The shaded area and black dashed line correspond to the same values as in the left panel. The blue dashed line indicates the mean log2 probability at each level of relative reward.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3713532&req=5

f1: Experimental task and learning data.The top panel provides a schematic of the experimental procedure. The lower left panel depicts mean probability of choosing a High option in a High/Low decision across participants on consecutive blocks of 6 decisions in each of the three experiments (error bars denote standard errors). The shaded area indicates probability below 80%, the threshold used to infer that participants learned to choose the High option, and the dashed line indicates 50%, the choice probability expected by chance. The lower right panel is a bubble plot of sensitivity to relative reward measured by the log2 probability of choosing a High option. As relative reward increased (the horizontal axis), more participants reliably chose the High choice in High/Low decisions. The size of each point is determined by the number of participants that obtained that value controlled for the number of participants in that experiment (i.e., probability density of that value within each experiment). The shaded area and black dashed line correspond to the same values as in the left panel. The blue dashed line indicates the mean log2 probability at each level of relative reward.
Mentions: Across three experiments, the ratio of the high-point reward to the low point reward was increased: Experiment 1 (7/5; n = 34), Experiment 2 (10/5; n = 37) and Experiment 3 (20/5; n = 55). Learning was faster and more reliable as the high-point value increased. On average, participants learned more quickly to choose the higher of the two options in the 20/5 experiment than in the other two experiments; after just 12 decisions, participants chose the high-value symbol on over 80 percent of High/Low decisions. By halfway through the experimental session (18 decisions), the high-value symbol was chosen reliably in High/Low decisions in all three experiments (see Fig. 1, bottom-left panel). The high-point value also affected the proportion of participants who learned to consistently choose the High options. The bottom-right panel of Fig. 1 provides the distribution of high-point/low-point choice ratio across participants in each experiment. The log2 of the ratio of the probability of choosing the high choice to the probability of choosing the low choice on any High/Low decision (i.e., log2(pHigh/pLow)) is employed as a measure of the degree to which the participant reliably chose High options in High/Low decisions (in the literature on basic learning principles, the log2 ratio is employed as an index of relative allocation of behavior to two independent responses and it has been shown to be sensitive to relative magnitude of reinforcement52.

Bottom Line: Within decisions, perceived alternatives compete until one is preferred.These decision spaces evolved through the experiments, as participants learned which options to choose.This visualisation approach provides a potential framework for the analysis of local dynamics in decision-making that could help mitigate both theoretical disputes and disparate empirical results.

View Article: PubMed Central - PubMed

Affiliation: School of Psychology, National University of Ireland, Galway, University Road, Galway, Ireland. denis.ohora@nuigalway.ie

ABSTRACT
Within decisions, perceived alternatives compete until one is preferred. Across decisions, the playing field on which these alternatives compete evolves to favor certain alternatives. Mouse cursor trajectories provide rich continuous information related to such cognitive processes during decision making. In three experiments, participants learned to choose symbols to earn points in a discrimination learning paradigm and the cursor trajectories of their responses were recorded. Decisions between two choices that earned equally high-point rewards exhibited far less competition than decisions between choices that earned equally low-point rewards. Using positional coordinates in the trajectories, it was possible to infer a potential field in which the choice locations occupied areas of minimal potential. These decision spaces evolved through the experiments, as participants learned which options to choose. This visualisation approach provides a potential framework for the analysis of local dynamics in decision-making that could help mitigate both theoretical disputes and disparate empirical results.

Show MeSH
Related in: MedlinePlus