VF - Visual Psychophysics - Week 1 Flashcards
How can the mechanisms of a black box be inferred?
Using an input, and observing the output.
Define threshold.
Smallest stimulus intensity required for detection or discrimination.
What is the threshold for wavelength discrimination?
> 2nm.
How are threshold and sensitivity related? Describe the formula.
They are inversely related
Threshold = 1 / Sensitivity
Why are threshold vs sensitivity curves often logged?
Taking the log allows representation of a large range of data in a compact form, allowing more practical representation, as well as more accurate interpolation.
Name the four methods of measuring threshold.
Method of adjustment
Method of limits
Method of constant stimuli
Staircase method (modified method of limits)
What are the two main errors when measuring threshold?
Error of habituation
Error of anticipation
Describe the method of constant stimuli.
Subject is randomly presented stimuli at different intensity levels. The data is collated and a psychometric function is constructed.
When measuring threshold, is there a steep step in which the subject never sees the stimulus to where they always see it?
No, it is instead a gradual change from consistently no, to consistently yes, and has an sigmoid curve.
The steepness at the inflection point depends on the sensitivity of the individual, with higher sensitivities resulting in a greater degree of steepness and vice versa.
On a psychometric curve, what is defined as the threshold?
The intensity at which 50% yes is reported.
How does modern threshold measurement compare to classical threshold theory?
Classical threshold theory attempted to deduce the neural threshold itself, and had steep steps from no to yes, where modern theory doesnt.
When presenting a subject with stimuli repeatedly, what is the classical threshold theory’s take on this vs modern theory? Explain why.
Classical theory asserts that repeated stimulation would always produce the same response.
This is false in modern theory due to the presence of noise - quantum fluctuations that mean the stimulus may not always be the same.
Similar noise will exist in the visual system as well, such as spontaneous chemical activity even in the absence of stimuli.
Is noise detected by the brain, or can it be ignored entirely?
No, signals are always detected alongside noise, and the brain must distinguish what is noise, and what is signal + noise.
Describe in detail the process by which the brain my interpret noise vs signal + noise. Explain the importance of the overlap, and the need for compromise. Mention why it isn’t feasible to have a 100% detection rate with no false positives, or why the intensity level should/shouldn’t be clear of the noise.
Consider noise to have a normal curve, with the mode at a given intensity. Signal + noise will have a similar normal curve, but with a mode higher than the noise normal curve. This results in an overlap between the two. The brain must select an intensity such that it recognises the signal, but with a low enough detection of noise.
Selecting an intensity clear of the noise may result in 0% false positives (ie. the mode), but doing so will mean a great portion of the curve is missed. Selecting an intensity that covers the entire signal + noise curve may result in a 100% detection rate, but will also mean a high false positive rate.
The brain compromises, and instead selects between the two, at around 85%.
Concerning noise vs signal + noise, what would the curve of hit rate vs false positive be called, and what does it look like? Explain what distance to the diagonal means, and whether a 100% hit rate and 0% false positive rate is possible.
It is hyperbolic, rapid increase, rapid plateau, called a receiver-operator curve. The steepness of the curve is related to the overlap between the normal curves of noise and signal + noise curves.
Higher degree of distance from the diagonal indicates greater separation of the normal curves.