Make your own free website on Tripod.com

Andrew's Online Journal

Evidence on the Questions of Dual Process Theory
Home
Research Papers
Non-Research Papers
How I Got Here
Places I like to Go on The Web
Contact Me

Andrew Clapper

University of North Carolina at Chapel Hill

Abstract

One interpretation of dual process theories proposes that the slower cognitive process is called the analytic system, and is more demanding to use because it requires the conscious use of rules and strategies.  If someone doesn't have the motivation or resources to use the analytic system, he or she must rely on the more instinctive, belief driven system called the heuristic system.  I propose there will be an interaction between cognitive load, working memory span, and conflict.   Based upon working memory score, subjects were divided into three groups after the experiment.  The subjects then evaluated the validity of logical syllogisms.  Two Easy problems and two Difficult problems were valid, and two of each problem were invalid.  Each subject was randomly assigned to one of three groups that varied in the cognitive load.  There was an interaction between cognitive load, working memory span, and conflict.  The fact that the different levels of working memory do not maintain position with regards to each other further demonstrates that there may not be a causal relationship between working memory and logical analysis.  However, it appears as though there are in fact separate systems being used in logical reasoning.  These findings are also consistent with those previously found.  Several possibilities exist for future exploration of research regarding dual processing theory, such as investigation of brain activity during similar experiments, as well as the usefulness of dual processing theory in explanations of orthodoxy.

 

 Introduction

People appear to be limited to a certain degree when it comes to logical puzzles, because they tend to make mistakes when they try to solve them (Gilovich et al., 2002)  When someone is patient and deliberate with a problem, they tend to make fewer mistakes (De Neys, 2006).  However, at times it seems as though solving a problem occurs so rapidly that it appears to have been done through instinct (De Neys, 2006).  In some instances, the rapid response to puzzles can result in a high error rate, if the puzzles are of a certain nature (De Neys, 2006).  Explanations for these kinds of errors are often called dual process theories (De Neys, 2006).

One interpretation of dual process theories proposes that the slower cognitive process is called the analytic system, and is more demanding to use because it requires the conscious use of rules and strategies.  People must recognize how to apply those rules and strategies, and they can only do so if they are sufficiently motivated and have the needed resources (De Neys, 2006).  If someone doesn't have the motivation or resources to use the analytic system, he or she must rely on the more instinctive, belief driven system called the heuristic system.  Therefore, a difference in working memory spans between individuals might result in different error rates in solving logic problems.  If a conflict exists between the heuristic and analytic systems when one is thinking about a logic problem, a higher working memory span will improve the likelihood that the person will reason the problem correctly.  If the person's working memory is unable to handle the amount of information it must process, he or she will be unable to use the analytic system, and will therefore be more likely to make a mistake (De Neys, 2006).   

De Neys (2006) gathered evidence related to his version of dual process theory by attempting to cause the working memory of subjects to be overloaded while they solved problems that required logical analysis.  The type of logic problem he used involved the evaluation of hypothetical syllogisms.  A hypothetical syllogism is valid when the third statement presented follows logically from the previous two given statements.  Subjects were asked to asses the validity of syllogisms presented.  Some of the statements conflicted with common knowledge, while others did not.  De Neys (2006) brought about a load on the working memory of the subjects by having them remember a visual pattern while they worked to solve the logic problems.  Some subjects had no memory task, others had to remember simple patterns, and the third group had to remember complex patterns.  De Neys (2006) believed that easy problems are solvable no matter the cognitive load because they are solved by using the heuristic system.  He proposed that cognitive load would have an effect on the ability to solve difficult problems, however, because they required the use of the analytic system.  A measurement of the working memory capacity of the subjects was obtained to allow him to assess the data he gathered (De Neys, 2006).

De Neys (2006) found that the storage task did not affect performance when the heuristic system was all that was needed to determine the correct response to a problem.  When the logical response conflicted with the instinctive response, the storage task was associated with decreased performance.  Evidence was that all individuals have both the analytic and heuristic systems, because performance on the logic problems decreased even in the low capacity groups, but performance when no conflict was present was unaffected (De Neys, 2006). 

If the procedure used by De Neys (2006) is closely approximated, a number of hypotheses can be tested.  I propose that a main effect of working memory (WM) will be observed, because individuals with the highest WM span will process the greatest amount of information over time.  Those with a high WM span will be followed by medium WM span, which in turn will come after those with a low WM span.  It is also hypothesized that there will be a main effect of cognitive load, because cognitive load will on average decrease performance the most when it is highest, followed by low cognitive load, followed by no cognitive load.  I propose there will be a main effect for conflict, because conflict will result in lower performance than no conflict.  I hypothesize that a WM Span x Conflict interaction will be observed, since the effect of WM Span on Throughput will be present only in the Conflict condition.  I also propose that a Cognitive Load x Conflict interaction will be present, because the effect of Cognitive Load on Throughput will only occur in the Conflict condition.  I hypothesize that there will be a Cognitive Load x WM Span interaction, since the effect of WM on Throughput will depend on the Conflict condition.  Lastly, I propose there will be a Cognitive Load x WM Span x Conflict interaction, because the interaction between Cognitive Load and WM Span will be observed only in the Conflict condition. 

 Method

The participants included 218 subjects.  Students enrolled in the Laboratory Research in Psychology class for the spring 2007 semester at the University of North Carolina at Chapel Hill were subjects.  The subjects reported to room 110 of Davie Hall at the University of North Carolina at Chapel Hill to participate in the study during their regularly scheduled class time.  The subjects followed prompts that appeared on the computer screens that gave directions and asked specific questions.

The first part of the study involved of a measure of working memory. Subjects evaluated arithmetic equations and were in the meantime asked to remember a series of words.  The number of words that the subject could remember was the measure of working memory.  Based upon working memory score, subjects were divided into three groups after the experiment.  The three groups were High, Medium, and Low working memory score. 

The second part of the study was meant to generate data to evaluate the main research questions.  The subjects evaluated the validity of logical syllogisms. The subjects had to solve eight syllogisms.  Four syllogisms were Easy, and four were Difficult.  The Easy syllogisms were either believable and valid or unbelievable and invalid.  For example, "Cats have tails.  Tigers are cats.  Therefore, tigers have tails."  The Difficult syllogisms were either believable and invalid or unbelievable and valid.  An example of a Difficult syllogism is "Bears have fur.  Gummy bears are a type of bear.  Therefore, gummy bears have fur."  Two Easy problems and two Difficult problems were valid, and two of each problem were invalid.  Each subject was randomly assigned to one of three groups that varied in the cognitive load.  For the High Load condition, the subject was asked to remember a complex pattern of dots while evaluating the syllogism.  Subjects had to remember a simple pattern of dots while solving the syllogism in the Low Load condition.  No secondary memory task was present in the No Load condition.  De Neys (2006) asked subjects to re-create the dot patterns in his experiment, but in this experiment the subjects chose the pattern they remembered from a choice of six patterns.

 

Results

No main effect of WM span (p = NS) was observed.  There was a main effect of Cognitive Load (F(1,210) = 5.1, p<.01), because the no Cognitive Load condition (x = .1229) showed higher Throughput than the low (x =.1065 ) and high Cognitive Load conditions (x = .1083). (See Table 1)  A main effect of Conflict (F(1,210) = 362.9, p<.001) was present because the Conflict group (x =.0829) showed lower Throughput than the No Conflict (x =.1406) (See Table 1).  There was no WM Span x Conflict interaction (p = NS).  However, there was a marginal Cognitive Load x Conflict interaction (F(2,210) = 2.97, p = .05) (See Figure 1), and a significant WM span x Cognitive Load interaction (F(2,210) = 2.7, p < .05) (See Figure 2).  There was also a Conflict x WM Span x Cognitive Load interaction (F(4,210) = 3.43, p = .01) (See Figure 3).

Discussion

The hypothesis that a main effect of working memory (WM) would be observed was not supported because individuals with the highest WM span did not process the information differently than those with medium or low WM span.  The hypothesis that there will be a main effect of cognitive load was supported, because the no Cognitive Load condition showed different Throughput than the high or low Cognitive Load conditions.  The hypothesis that there would be a main effect for conflict was supported because the Conflict group showed different Throughput than the No Conflict.  The hypothesis that a WM Span x Conflict interaction would be observed was not supported, since the effect of WM Span on Throughput was present in both conditions.  The hypothesis that a Cognitive Load x Conflict interaction will be present was supported, because the effect of Cognitive Load on Throughput will occurred differently between No Conflict and Conflict.  The hypothesis that there would be a Cognitive Load x WM Span interaction was supported, since the effect of WM on Throughput depended on the Conflict condition.  Lastly, the hypothesis that there will be a Cognitive Load x WM Span x Conflict interaction was supported, because the interaction between Cognitive Load and WM Span was different than the interaction between WM Span and Conflict.

The fact that there was no main effect of WM casts doubt on the idea that those with high WM will have more resources to provide the analytic system to use.  However, this finding is consistent with what De Neys (2006) found.  These findings suggest that WM level does not necessarily predict how well someone uses the analytic system, or how well they might do with logical analysis.  Also, the fact that the different levels of working memory do not maintain position with regards to each other further demonstrates that there may not be a causal relationship between WM and logical analysis.  However, since Throughput was in fact affected by Cognitive Load and Conflict, it appears as though there are in fact separate systems being used in logical reasoning.  These findings are also consistent with those found by De Neys (2006). 

A number of improvements may help the study yield higher quality data to be analyzed.  A larger number of logic problems may result in more consistent trends in the data, although subject fatigue might then become a risk.  Also, the study can be improved by using different sampling methods.  For example, this study used an ad hoc sample of psychology students at one particular school, and in order to be able to generalize to the rest of the human population, either a more random sample should be obtained or it should be demonstrated that the ad hoc sample does not differ from the general population with respect to cognitive functioning.  It may also be case that the study needs a larger sample in order to yield a normal distribution of data. 

Several possibilities exist for future exploration of research regarding dual processing theory.  Perhaps if the brain activity and blood flow are measured while the study is conducted, researchers may gain insight into what regions in the brain are used during the cognitive processes under investigation.  Such measurements might also provide evidence that either supports or contradicts dual processing theory itself.  It is possible that dual processing theory may help explain the phenomena of orthodoxy, and the idea deserves further investigation.  Perhaps culturally orthodoxical ideas can be used in the syllogisms and be compared to non orthodoxical ideas as one way of investigating the possibility of dual processing and orthodoxy.  Also, in a similar line of reasoning, perhaps ideas involving cultural norms could be tested against each other in order to investigate the role of environment on what De Neys (2006) calls the belief based heuristic system. 

 References

De Neys, W. (2006). Dual processing in reasoning: two systems but one reasoner.  Psychological Science, 17, 428-433.

Gilovich, T., Griffin, D., & Kahneman, D. (Eds.) (2002). Heuristics and biases: The psychology of intuitive judgment. New York: Cambridge University Press.

 
Appendix: Table 1
Condition Throughput
High Load 0.1083
Low Load 0.1065
No Load 0.1229
High WM 0.1148
Medium WM 0.1103
Low WM 0.1108
Conflict 0.0829
No Conflict 0.1406

 

 

project2figure1.jpg

project2figure2.jpg

Figure 3

project2figure3.jpg

project2figure3b.jpg

Welcome to my college adventures.