Task 2 - Signal Detection Theory Flashcards
Signal detection theory (SDT)
A framework for measuring decision-making under uncertainty by analyzing the ability to distinguish signals from noise, considering both sensitivity and response bias
Internal noise
Variability in neural responses caused by biological and physiological factors
External noise
Variability in the stimulus itself due to environmental factors
Four possible outcomes in SDT
- Hit (signal present and subject says yes)
- Miss (signal present and subject says no)
- False alarm (signal absent and subject says yes)
- Correct rejections (signal absent and subject says no)
The role of the criterion in SDT
The decision threshold a person sets to determine whether they perceive a signal.
* A liberal criterion → Lower threshold for saying “yes” → More hits but also more false alarms.
* A conservative criterion → Higher threshold for saying “yes” → Fewer false alarms but also more misses
The choice of criterion depends on bias
The Receiver Operating Characteristic (ROC)
We can describe the full range of a persons options in the ROC curve
* The ROC curve plots the false alrm rate on the x axis and the hit rate on the y axis
* If the criterion is moved lower, both hit rate and false alarm rate decrease
d’
Discriminability index: The higher the further apart the peaks are in the probability density function
So: Depends on both the strenght of the signal and the variability of the noise
d’ = seperation/spread
Probability density function
- Shows two curves for noise and signal + noise. The higher the overlap of curves, the lower the discriminability
- Plots probability that a perception is caused by N or by S+N against perceptual effect
Is there an absolute threshold when a person goes from hearing something to not hearing something?
SDT implies that this is not the case and the threshold can be changed by the persons response criterion
Signal and Noise
SDT says that individual always has to differentiate between signal and noise when making a decision:
Signal: Stimulus presented to the subject
Noise: All the other stimuli in the environment
Beta
- The ratio of neural activity produced by signal and noise at the criterion
- Increases over 1 when the criterion is shifted to the right, decreases below 1 when shifted to the left
- SDT is able to tell you where beta should be based on costs and payoffs of the four possible outcomes
Sluggish beta
Studies showed that participants demonstrate a sluggish beta, meaning they are less conservative than they should be if the ideal beta is high, and less liberal than they should be if the ideal beta is low
The role of signal strenght
If the signal is stronger, the probability density function of the signal+noise trials shifts to the right, away from the function for noise (peaks are further seperated)
Optimal beta
Depends on both the likelihood of observing signal vs. noise (P(N)/P(S)) and the payoffs (costs and benefits of the four outcomes)
Payoffs
If payoffs higher for hits or costs higher for misses - low criterion
If payoffs higher for correct rejections or costs higher for false alarms - high criterion