Psychophysics Flashcards
What is the psychophysical function?
‘If I put twice as much salt in my soup, is it twice as salty?’
NOT TRUE, psychological intensity varies to actual intensity (grows exponentially).
Define the Just-Noticeable Difference
The smallest difference you can detect in a change in stimulus.
Changes depending on the magnitude of the stimulus.
Weber’s Law
The JND is a constant proportion of the stimulus intensity, how much you have to change a stimulus for it to be noticeable. Specific for each sense: -Pitch = 1/333 -Brightness = 1/60 -Loudness = 1/10 -Taste = 1/5
Outline perceptual thresholds
When is a stimulus just perceivable? or when is a difference perceivable?
No absolute threshold for perception, neural machinery is noisy and lacks consistency.
What are the three threshold finding methods?
- Method of constant stimuli
- Method of adjustment
- Method of limits
ALL HAVE RESPONDER BIAS
Outline Method of Constant Stimuli
- Non-sequential choices
- Slower, no order effects
Outline Method of Adjustment
- PP adjusts stimulus
- Quick
Method of Limits
- Uses ascending and descending trials
- Can predict when stimulus is close.
Outline Signal Detection Theory
- You make a trade-off between these types of errors depending on your threshold, trading off false alarms and misses.
- If you reduce your false alarms, you might get an increase in misses
- Where you place your threshold depends on the costs and biases of errors
Example of Signal Detection Theory
- Phantom phone buzz, because you perceive messages as important so don’t really mind if its a false alarm because you don’t want misses.
- Your perceptual system through experience lowers your criteria/threshold
Outline ROC Curves
- Graph the number of false positives on the x-axis and the number of hits on the y-axis, put a point where the individual hits.
- If there’s fewer hits, consequently fewer false alarms (A), with more hits there will be more false alarms (B).
- A and B have the same sensitivity but just have different biases.
What is d’ ?
A measure of sensitivity
d’ = Z(fa) - Z(hits)
What is a Z-score?
Distance in standard deviations from the extended score.
Converts a number into another number
In –> a proportion between 0 and 1
Out –> a number between -infinity and +infinity
How can you measure accuracy in detection and discrimination where bias can’t contribute?
Use forced choice decisions e.g. is the stimulus on the left or right?