Poster 6

Diagnosing Human Flawed Intuition of Randomness; Bayes vs Fisher compared

Kimberly Rodriguez and other Students in Robert Becklen's Research Methods and Data Analysis sections

The inability of humans to accurately intuit randomness has long been a topic of study in the fields of cognition and more recently neuroscience. In this project we show the power of multivariate approaches to the identification of flawed intuitions by specifically comparing Baysean inspired model building vs the traditional insistence of the ""null hypothesis"" constraint by famed statistician Ronald Fisher. We show that in practical situations the use of prior knowledge in the assessment of departures from randomness outperforms the Fisherian approach when assessed against external validation data sets never encountered during the model building stages. The insights offered by this approach also has implications for recent efforts to identify cognitive and neurological deficits through the assessment of algorithmic complexity of human generated randomness grounded in information-science.

(Click upper right corner of image below to enlarge, also note there are multiple slides so click for next slide)