Vertical-Horizontal Illusion

Download Report

Transcript Vertical-Horizontal Illusion

Longer or Shorter?
An examination of line length discrimination using
manipulated feedback
Amanda Hostiuc, Alicia Kim, Melanie Laking, and Matt Pachai
Outline
1) First steps and hypothesis generation
– The Peter Jansen paradigm
– Failures and revisions
– The Vertical-Horizontal Illusion
2) Experimental Design
3) Results
4) Conclusions
The Peter Jansen Paradigm
• The original task flashed two lines on screen
simultaneously and asked if the second line
was longer or shorter
• This experiment served as the starting point in
our research program
The Original Motivation
• What is the effect of attention on our line
length acuity in the Peter Jansen paradigm?
• Could attentional cueing lead to increased
performance on a length discrimination task?
• Would an invalid cue decrease performance?
A Snag in the Literature
• A disturbing picture
began to emerge
• Many articles described
the “well known fact”
that vertical lines are
perceived as longer
than horizontal lines
The Vertical-Horizontal Illusion
• When a horizontal line and a vertical line of the same
length are presented together, the vertical line is
perceived as on average 10% longer
Craven (1993)
• Systematically examining the perception of
The New Design
• We decided it was unlikely that cuing would
cause a significant change in like acuity if our
perceptions are already biased
• But what if we could change the perception of
the illusion?
• How might you go about designing an
experiment to manipulate
Programming a Pilot Run
• We decided to program our experiment from
scratch in Matlab due to prior experience
• After extensive tweaking, Amanda and
Melanie ran as control subjects
• The result
 NO vertical-horizontal illusion was present!
Back to the Drawing Board
• After our initial failure, we began to scour
journal articles to find an ex
Vertical-Horizontal Illusion
• Vertical-Horizontal Illusion (V-H)
– Perception of vertical lines as longer than
horizontal
Purpose
• Increase or eliminate vertical-horizontal
illusion
Idea Generation (edit)
• Consecutive lines
• Cues
– Same/different
– Shorter/longer
• Feedback
– Correct
– biased
Herzog and Fahle, 1999 --Vernier Task
Hypothesis
• Valid feedback
– Decrease V-H
• Biased feedback
– Increase V-H
Method
• Conditions
– Reduction: valid feedback
– Control: no feedback
– Strengthened: biased feedback
Method
• Blocks (Phases)
– Block 1: establish baseline threshold
– Block 2: learning trials
– Block 3: examine the effect of learning trials
compared to baseline threshold
Design
Block 1
Block 2
Block 3
Reduction
No
feedback
Valid
feedback
No
feedback
Control
No
feedback
No
feedback
No
feedback
Strengthened
No
feedback
Biased
feedback
No
feedback
Method
Method
Method
• Subjects
– 8 subjects
• Stimuli
– Vertical/horizontal lines between -9 to +3 pixels
Method
• Measure
– Accuracy of the responses
• Trials
– 10 practice trials
– 240 trials for each block
– 20 trials per length per block
– Total of 720 trials
Method
• Procedure
– Fixation cross
– Mask
– Stimulus
– Response: left or right
+
+
Result
Phase 1
Phase 3
Difference
Subject 1
Subject 2
-3.2651
-3.1660
-2.66190
-2.5070
1.6032
0.6590
Subject 3
Subject 4
Subject 5
-3.2703
-3.8623
-3.5175
-4.0302
-4.4380
-3.5103
-0.7593
-0.5757
0.0072
Subject 6
Subject 7
Subject 8
-3.3370
-2.1695
-5.6957
-4.0723
-1.8156
-3.8027
-0.7353
0.3539
1.8930
Condition 1
Condition 2
Condition 3
Results (Condition 1)
Results (Condition2)
Results (Condition 3)
Results
• Replicated the v-h illusion
• Inconclusive results in reducing/
strengthening the illusion with valid/biased
feedback
Discussion
• Finding 1: Vertical-horizontal illusion was
replicated
– Craven (1993)
Discussion
Finding 2:
• Unable to decrease illusion using valid
feedback
Finding 3:
• Unable to increase illusion using biased
feedback
Discussion
Source of error
– Number of trials
– Number of N
Discussion
Suggestions for future research
– More time to train/learn
– More trials for Block 2
– Better incentive