Many fields of human endeavour require reaching conclusions with small
data samples. Unlike traditional data mining algorithms, humans make inferences
on small samples, forget the data, and then change their conclusions as new data
are obtained. Humans thus learn across many different data sets with small memory
IHMC researchers use a variety of techniques to study human data-mining.
Modeling of dynamical, low-memory associative learning defines limits on
learning procedures. Systems in which complex functions and causes underlie
observed variables are analyzed statistically. Characterization and testing
of algorithms for learning causal relations contribute to the development of
new algorithms. We also study strategies young children use in learning about
cause and effect.
Understanding such processes is important not only for cognitive psychology
but also for many circumstances where automated learning methods are needed, for
example in space exploration where robotic explorers may have very limited
memory capacities, and in any circumstance in which related data are collected
in separate databases.