Title: Modeling Episodic Memory
1Modeling Episodic Memory
- PROJECT Using a web based modeling environment
called NetLogo, I attempted to create a working
model of the migration of objects and information
across different episodic memories using an
alternate computing system known as Cellular
Automata. - EPISODIC MEMORY An episodic memory is a memory
of a specific event like a trip to the
supermarket or dinner at a restaurant. A
migration error occurs when an object, inference,
or piece of a single episodic memory moves to
another episodic memory. - Experiments preformed by the SLU psychology
department have provided empirical data showing
that migration errors occur more frequently with
memories of the same schema, or setting. - CELLULAR AUTOMATA A cellular automaton is a
computing system whose behavior is dependant upon
the local relations of the automata within it.
They are often used to simplify complex
biological systems. The best way to think of a
cellular automaton to visualize it as a 2-D gird
of cells. In its simplest form a cell is either
on/alive or off/dead. Each cell acts based on
whats going on in the cells around it. - DATA My dataset consists of 20 entries 10
restaurant and 10 supermarket settings. Each one
consists of 14 characteristics and a set of
objects. There are 3 different types of
characteristics. Those only possessed by
restaurants, those only possessed by
supermarkets, and those possessed by both. All
of this data was provided by Dr. Sharon Hannigan
of the SLU Psychology department.
Migration Zapper
Aged Memory Zone
Episodic Memory
Fresh Memory Zone
NetLogo is a programmable modeling environment.
It allows you to create simple models with ease,
access an extensive library of pre-made models,
or to create complex models by programming every
aspect using its built in language. Each pink
icon represents a distinct episodic memory. The
older memories are farther from the center and
the and most recently recalled memories are close
to the center. The yellow lightning bolt is a
zapper. It acts as a mutation variable. It
moves randomly while favoring the outer edges.
Any memory it hits has a chance of being mutated.
Sample Memory Data
RESULTS My final model was able to recreate
migration errors similar to those found by the
psychology departments experiment. Migration
errors caused objects and information to move
between memories and caused the model accurately
fail to recall different memories correctly. I
was unable to finish the final step of my
project, incorporating the physical proximity of
one memory to another affecting how likely the
two are to manifest a migration error. This does
leave extra avenues of research for future
extensions of this modeling project.
Researcher Ben Case Related Researcher Adam
Britt Advisor Dr. Ed Harcout Cognition Advisor
Dr. Sharon Hannigan