Kool, W., McGuire, J., Rosen, Z. and Botvinick M. (2010). Decision Making and the Avoidance of Cognitive Demand . Journal of Experimental Psychology: General, 1-18.
@Article{KoolMcguireRosenBotvinick2010,
author = {Kool, W. and McGuire, J. and Rosen, Z. and Botvinick M.},
title = {Decision Making and the Avoidance of Cognitive Demand},
journal = {Journal of Experimental Psychology},
year = {2010},
pages = {1-18}
}
Author of the summary: Derrick Matthew Buchanan, 2012, dmab98@hotmail.com
Cite this paper for:
The full paper can be found online at http://www.princeton.edu/~matthewb/Publications/KoolETAL_JEPG_inPress.pdf
The law of less work: "If two or more behavioural sequences, each involving a different amount of energy consumption or work, have been equally well reinforced an equal number of times, the organism will gradually learn to choose the less laborious behaviour sequence leading to the attainment of the reinforcing state of affairs. [p.1]
In this paper, the authors conduct experiments to extend Hull's, already tested and validated, law of less work, to the law of least mental effort. It was already assumed that Hull's law was extended to mental effort, however the authors conduct 6 experimental tests in order to empirically validate the extension of Hull's law from physical to mental effort.
"There is also direct evidence that human agents offload control demands when possible, relying on information in the perceptual environment rather than internal working memory or cognitive control representations." [p.2]
Current Experiments:
All 6 of the experiments conducted are built around the use of demand selection tasks (DSTs). Each of the experiments progressively controls for different explanations to the hypothesis, in order to eliminate them and ensure the validity of their hypothesis. The author's overall hypothesis across the experiments was that "the participants would develop a tendency to select the course of action associated with the least cognitive demand." [3]
Experiment 1: [p.3-5]
43 subjects from University of Pennsylvania (18-26 years of age; 25 women, 18 men).
500 trials of DST conducted for each participant.
2 digital decks provided; low demand deck and high demand deck: The subjects are blind to the level of cognitive demand built into the decks.
Participants are free to choose which deck they want to use at the beginning of each trial.
Two- way repeated measures analysis of variance (ANOVA) was conducted for whether participants repeated the same deck or switched the deck at the beginning of each of the 500 trials.
Results for deck choice: from 500 trials, mean proportion of trials that low demand was selected was 0.68 (SD= 0.24). [4]
To analyze and validate error rates between high and low demand decks they were compared using Wilcoxon signed- rank task.
Results for error rates: mean error for low demand deck is 1.73%, and mean error rate for high demand deck is 2.58%. [4]
This is a significant difference (p=.054).
Overall results: showed 36 subjects (84%) selected the low demand deck more often than the high demand deck.
Discussion: "Results appear consistent with a law of least mental effort, the idea that, all else being equal, actions tend to be selected to minimize cognitive demand." [4]
Experiment 2: [p. 5-7]
"If demand avoidance is a ubiquitous characteristic of behaviour, then it should occur under demand manipulations different from those used in the first experiment." [5]
24 subjects from Princeton University (18-40 years of age; 14 women, 10 men).
Subjects completed 6 trial blocks of A-X task, and have to choose one of two alternatives at the beginning of each block.
Subjects were again presented with high and low demand alternatives in a DST.
High demand selection probed ABABAB or BABABA (requires 5 context shifts), low demand selection probed AAABBB or BBBAAA (requires one context shift).
The mean accuracy rate for the low demand blocks were 0.95 (SD=0.05). Mean accuracy rate for high demand blocks were 0.90 (SD=0.08). These are significantly different (t(23)=5.49, P<.01). [6]
Low demand option selected at mean rate of 0.64 (SD=0.27). Analysed through Wilcoxon signed-rank test found this proportion to differ significantly from
chance (p=.03). [6]
Discussion: By replicating the demand avoidance bias in this new context it provides evidence that "people avoid high-demand courses of action systematically
rather than, for instance, merely growing less likely to repeat their last motor response when the task is more demanding." [6]
Experiment 3: [p.7-8]
This experiment is controlling for the small minority who seemed to show preference to high demand tasks. The prediction is that if they can reduce the impact of arbitrary cue-related preferences, then the frequency of biases toward high demand should also reduce.
Given two choice cues differing in levels of cognitive demand, tested in multiple runs with changing locations of choice cues each run.
37 subjects from Princeton University (18-30 years of age; 21 women, 16 men).
75 trials in each of the 8 runs of experiment (600 total).
Has the same set up as Experiment 1, but the position of cues changed from run to run.
“The DST showed internal consistency in assessing the bias of individual subjects to avoid cognitive demand”: Cronbach's alpha =.85. [8]
Discussion: Goal of this experiment was to reduce influence of demand-independent cue or response preferences in DST, and after conducting the experiment it was found that no subjects exhibited a systematic preference for the high demand, but it was rather more likely that the subjects who seemed to seek high cognitive demand were actually guided by demand irrelevant facts. [8]
Experiment 4: [p.8-9]
Experiment 4 attempts to find out if the tendency to avoid cognitive demand applies across different varieties of cognitive demand.
This task subjects had to perform 2 digit mental subtraction; with a manipulation on whether or not the problem involved carrying.
Since carry options increase cognitive demand (Hitch, 1978), they hypothesized that participants would avoid solving the problems with carrying.
16 subjects from Princeton University (18-22 years of age, 10 women, 6 men).
Task is to verify if the answer to the presented subtraction problems is correct or not: 8 runs of task, 5 min each.
Low demand subtraction required no carrying, high demand subtraction required carrying.
Mean RTs were 1207ms for low demand trials, and 2026ms for high demand trials. These differed significantly (Wilcoxon signed-rank. p<.01). [8]
Mean low demand selection rate was 0.73. Differed significantly (Wilcoxon signed-rank. p<.01). [8]
Another crucial result was that no participant showed a significant bias in the opposite of expected direction. "DST showed high internal consistency (Chronbach's alpha=.93)." [8]
Discussion: The results from this experiment supplement the results from Experiment 2 by "extending the law of least mental effort beyond just task switching." [9]
Experiment 5: [p.9-10]
This experiment attempts to control and test for individual differences.
Hypothesized that meaningful individual differences in task-switching abilities may be present among subjects.
Further prediction is that "if variation in demand avoidance were related to ability, then individuals showing a larger switch cost in the preliminary period would be expected to go on to show higher levels of avoidance." [9]
19 subjects from Princeton University (18-27 years of age; 11 women, 8 men).
Used DST similar to experiment 3.
Involved 8 runs with 75 trials each.
The evaluate the effect of individual differences in ability on preference, they tested for the correlation of total low demand preference rates with RT switch costs from preliminary choice free block. [10]
Mean low demand selection rate was 0.67. (range= 0.45 to 0.95) Differed significantly from 0.5 (Wilcoxon signed-rank p<.01). Again DST showed high internal consistency (Cronbach's alpha=.91). [10]
7 subjects showed significance for demand avoidance, and 12 subjects did not reach statistical significance but still showed aversion of high demand tasks.
Discussion: Overall the data reflected the expected predictions. That "individuals whose resources for controlled information processing are more heavily taxed by a given task should avoid that task relatively strongly." [10]
Experiment 6 a&b [10-14]
Experiment 6 was broken into a&b (combined into one section since they are testing for the same prediction).
This experiment attempts to decouple RTs from the time required to complete task objectives. This is because other models have proposed that our decision strategies are made to reduce response time rather than cognitive demand. (Bogacz et al 2006)
This experiment uses new DST.
62 subjects from Princeton University and 22 subjects from Leiden University (17-33 years of age; 50 women, 12 men).
57 participants played 110 games; remaining participants played variable number of games for duration of 30 min.
No statistically significant difference between Leiden and Princeton groups, or between the groups with fixed time duration versus fixed game number. [12]
Discussion: After conducting this experiment they were able to conclude that "absent compensating incentives, people tend to avoid cognitive demand." They were able to make this conclusion because during the performance of this experiment participants would tend to avoid switching task strategies, despite the fact that some circumstances made switching strategies the fastest way to complete the task. [13]
Conclusion from 6 experiments is that they all collectively appear to provide evidence for the law of least mental effort.