Limits...
Search, Memory, and Choice Error: An Experiment.

Sanjurjo A - PLoS ONE (2015)

Bottom Line: Despite standard models of choice treating such variations in order of acquisition as irrelevant, lower predicted memory load search orders are found to lead to substantially fewer choice errors.Further, as cognitive load is induced endogenously (within the task), and found to affect choice behavior, this result contributes to the cognitive load literature (in which load is induced exogenously), as well as the cognitive ability literature (in which cognitive ability is measured in a separate task).In addition, while the information overload literature has focused on the detrimental effects of the quantity of information on choice, this result suggests that, holding quantity constant, the order that information is observed in is an essential determinant of choice failure.

View Article: PubMed Central - PubMed

Affiliation: Fundamentos del Análisis Económico, Universidad de Alicante, Alicante, Spain.

ABSTRACT
Multiple attribute search is a central feature of economic life: we consider much more than price when purchasing a home, and more than wage when choosing a job. An experiment is conducted in order to explore the effects of cognitive limitations on choice in these rich settings, in accordance with the predictions of a new model of search memory load. In each task, subjects are made to search the same information in one of two orders, which differ in predicted memory load. Despite standard models of choice treating such variations in order of acquisition as irrelevant, lower predicted memory load search orders are found to lead to substantially fewer choice errors. An implication of the result for search behavior, more generally, is that in order to reduce memory load (thus choice error) a limited memory searcher ought to deviate from the search path of an unlimited memory searcher in predictable ways-a mechanism that can explain the systematic deviations from optimal sequential search that have recently been discovered in peoples' behavior. Further, as cognitive load is induced endogenously (within the task), and found to affect choice behavior, this result contributes to the cognitive load literature (in which load is induced exogenously), as well as the cognitive ability literature (in which cognitive ability is measured in a separate task). In addition, while the information overload literature has focused on the detrimental effects of the quantity of information on choice, this result suggests that, holding quantity constant, the order that information is observed in is an essential determinant of choice failure.

No MeSH data available.


Relative frequency of correct choices as search order and matrix size vary.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4487248&req=5

pone.0126508.g005: Relative frequency of correct choices as search order and matrix size vary.

Mentions: Fig 5 reports relative frequency of correct choices, averaged across subjects (with standard errors), as predicted WML varies with search order and matrix size. Both hypotheses stated in the second section are tested using a repeated measures ANOVA, with two repeated factors: search order and matrix size. H1 consists of three tests: one test of the simple effect of search order for each matrix size. As is clear from the hypotheses section, for matrix size 2 the test is two-sided, whereas for matrix sizes 3 and 4 the tests are one-sided. In order to address the multiple comparison issue the Bonferroni correction is used, as described below. In the 2×2 matrices of attributes in which AL and AT searchers experience the same predicted amount of maximum WML, AL searchers are shown to make the correct choice in 94.66% of tasks while AT searchers make the correct choice in 92.93% of tasks; this difference is not statistically significant (0.0172 [0.0214], p = .423), where the numbers enclosed in parentheses are (average difference [standard deviation], and p-value]. Meanwhile, in the 3×3 matrix, where AL searchers are predicted to experience less maximum WML than AT searchers (3 vs. 5, respectively) AL searchers make the correct choice in 87.59% of tasks, while AT searchers make the correct choice in 77.93% of tasks; this difference is statistically significant (0.0966 [0.0214], p < .001). Finally, in the 4×4 matrix, where AL searchers are predicted to experience much less maximum WML than AT searchers (3 vs. 7, respectively) AL searchers make the correct choice in 76.63% of tasks, while AT searchers make the correct choice in 48.66% of tasks; this difference is statistically significant as well (0.2797 [0.0214], p < .001).


Search, Memory, and Choice Error: An Experiment.

Sanjurjo A - PLoS ONE (2015)

Relative frequency of correct choices as search order and matrix size vary.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4487248&req=5

pone.0126508.g005: Relative frequency of correct choices as search order and matrix size vary.
Mentions: Fig 5 reports relative frequency of correct choices, averaged across subjects (with standard errors), as predicted WML varies with search order and matrix size. Both hypotheses stated in the second section are tested using a repeated measures ANOVA, with two repeated factors: search order and matrix size. H1 consists of three tests: one test of the simple effect of search order for each matrix size. As is clear from the hypotheses section, for matrix size 2 the test is two-sided, whereas for matrix sizes 3 and 4 the tests are one-sided. In order to address the multiple comparison issue the Bonferroni correction is used, as described below. In the 2×2 matrices of attributes in which AL and AT searchers experience the same predicted amount of maximum WML, AL searchers are shown to make the correct choice in 94.66% of tasks while AT searchers make the correct choice in 92.93% of tasks; this difference is not statistically significant (0.0172 [0.0214], p = .423), where the numbers enclosed in parentheses are (average difference [standard deviation], and p-value]. Meanwhile, in the 3×3 matrix, where AL searchers are predicted to experience less maximum WML than AT searchers (3 vs. 5, respectively) AL searchers make the correct choice in 87.59% of tasks, while AT searchers make the correct choice in 77.93% of tasks; this difference is statistically significant (0.0966 [0.0214], p < .001). Finally, in the 4×4 matrix, where AL searchers are predicted to experience much less maximum WML than AT searchers (3 vs. 7, respectively) AL searchers make the correct choice in 76.63% of tasks, while AT searchers make the correct choice in 48.66% of tasks; this difference is statistically significant as well (0.2797 [0.0214], p < .001).

Bottom Line: Despite standard models of choice treating such variations in order of acquisition as irrelevant, lower predicted memory load search orders are found to lead to substantially fewer choice errors.Further, as cognitive load is induced endogenously (within the task), and found to affect choice behavior, this result contributes to the cognitive load literature (in which load is induced exogenously), as well as the cognitive ability literature (in which cognitive ability is measured in a separate task).In addition, while the information overload literature has focused on the detrimental effects of the quantity of information on choice, this result suggests that, holding quantity constant, the order that information is observed in is an essential determinant of choice failure.

View Article: PubMed Central - PubMed

Affiliation: Fundamentos del Análisis Económico, Universidad de Alicante, Alicante, Spain.

ABSTRACT
Multiple attribute search is a central feature of economic life: we consider much more than price when purchasing a home, and more than wage when choosing a job. An experiment is conducted in order to explore the effects of cognitive limitations on choice in these rich settings, in accordance with the predictions of a new model of search memory load. In each task, subjects are made to search the same information in one of two orders, which differ in predicted memory load. Despite standard models of choice treating such variations in order of acquisition as irrelevant, lower predicted memory load search orders are found to lead to substantially fewer choice errors. An implication of the result for search behavior, more generally, is that in order to reduce memory load (thus choice error) a limited memory searcher ought to deviate from the search path of an unlimited memory searcher in predictable ways-a mechanism that can explain the systematic deviations from optimal sequential search that have recently been discovered in peoples' behavior. Further, as cognitive load is induced endogenously (within the task), and found to affect choice behavior, this result contributes to the cognitive load literature (in which load is induced exogenously), as well as the cognitive ability literature (in which cognitive ability is measured in a separate task). In addition, while the information overload literature has focused on the detrimental effects of the quantity of information on choice, this result suggests that, holding quantity constant, the order that information is observed in is an essential determinant of choice failure.

No MeSH data available.