How to investigate Perception & Cognition n Ask your subjects (Introspectionism) n Look at S-R patterns (Behaviorism) n Infer mental processes (Cognitive Psychology) –from S-R patterns (Reaction Time, Accuracy) –from neural patterns (cognitive neuroscience) Last Class we ended with

Cognitive Psychology Study stimulus-response relations to infer the underlying mental processes. The contents of the mind CAN be studied scientifically. How? Three models: The ‘black box’, the ‘jukebox’, the ‘mind box’ Representations Stimulus Response Information Processing

The black box Stimulus (input) Response (output) 1. Observe the Input/output relation 2 create a ‘model’ (abstract representation) to account for the I/O - requires a little bit of background knowledge (e.g., naïve physics, naïve biology), - It is a ‘functional’ model (how the box functions)

The black box: A ‘model’ Stimulus (input) Response (output) - Is ‘A’ stick connected to the brown stick? - Can we run an experiment to find out? Force (push) Pull A -vary input & measure output, from that I/O pattern make: - inferences about the internal constructs (i.e. create a functional model)

The jukebox ABCDEFGHIJKLMN Input (selection) Output (a particular song) As with the ‘black box’, we need to infer how it works (a functional model). We do this from the I/O pattern, and a little bit of background knowledge. ANY GUESSES?

A B How can we test this model? - Assume that selector always starts from top - Measure how long it takes to start playing selector Stack of disks Turntable

Graph data

Move selector arm to top of stack Is record selected one? Move selector to next record no Put record on turntable yes Functional Model of the ‘jukebox’ A B selector Stack of disks -A hardware - A ‘functional’ architecture (this is what Cog Psy studies)

The mind box n The same approach to study the functional architecture of the jukebox, can be applied to the study of the functional architecture of the mind. n Create an internal model by looking at the I/O pattern –Input (independent variable) –Output (dependent variable): usually Reaction Time (RT) n Example: Sternberg search (web).web –A set of letters will appear briefly on the screen for you to study –The letters will disappear –3 seconds later one probe item will appear. –Your task is to decide whether the probe item was in the initial list. Say ‘yes’ if the probe was in the list, ‘no’ if it wasn’t. READY?

P Z T A C H H Memory set (t 1 ) - Set size can vary Probe (t 2 ): -Present in set, or -Absent in set delay P Z T A C H

XTNBD Memory system Y Y Create an input representation X Compare the two representations Y match respond “yes” no match last memory rep? yes respond “no” no X Retrieve a representation from memory Sequential Model: Predictions - Larger memory set  slower RT -‘yes’ trials  faster RT -‘yes’ trials  shallower slope

Memory system Y Y Create an input representation XTNBD Y Compare to all representations simultaneously match respond “yes” no match respond “no” Parallel Model: Predictions - Larger memory set  same RT -‘yes’ trials  same RT -But caveats exist!

Contrasting the two models n I.V.s (& levels): –?? n D.V.:?

X T

B

N B D

N

X T N B D

T

Sternberg (1967)