Top-Down Processing Psych 240

Signal detection theory

a theory predicting how and when we detect the presence of a faint stimulus amid background stimulation. Assumes there is no single absolute threshold and that detection depends partly on a person's experience, expectations, motivation, and alertness.

signal

something in the environment you are trying to detect

noise

things/ disturbances in environment other than the signal

sensitivity

how easy/ difficult it is to distinguish signal from noise; function of both the distinction and how good your perceptual apparatus is. The lower the sensitivity the worse the detection. High sensitivity = good at distinguishing a real signal from a fake one

bias

The threshold you choose for how strong the signal must be for you to say its there. Determined by expectations or payoffs and how high a threshold you have

hits

there is a signal and you correctly detect it

misses

there is a signal but you fail to detect it

correct rejections

there is no signal and you correctly say there was no signal

false alarms

there is no signal nut you say that there was a signal - Error

bias/payoff effects

Yes Bias - low threshold when you say yes whenever you think there is a possibility of the signal being there

No Bias - having a high threshold, when you say no whenever there is a possibility of being wrong

Context effects

When perception of an object is affected by its context/ environment; Surrounding context creates top down influences on our perception of different stimuli

subjective contours

Perceived contours that do not exist physically. We tend to complete figures that have gaps in them by perceiving a contour as continuing along its original path.

Word superiority effect

you are better at detecting the letters when they are IN words, then when they are by themselves

pseudoword superiority effect

letters are easier to identify when in pronounceable nonwords than random strings of letters

interactive activation model

Explains word superiority effect; a theory proposing that both feature knowledge and word knowledge combine to provide information about the identity of letters in a word

activation

exciting a neuron/ causing another neuron to fire

inhibition

A synaptic message that prevents a recipient neuron from firing.

interactive activation model: Feature detectors level

What makes that neuron fire of which then send activation up to the letter level

- Send excitatory signals to features present in a letter

- Send inhibitory signals to features not present in the letter

interactive activation model: Letter Level

Respond to whole letters based on the features that are detected to be present in a specific letter

- Sends activation up to the word level

interactive activation model: Word Level

Creates the word; Inhibits the wrong words and Excites the correct words

- Can create top down processing to letter level → once we start getting some evidence for the word unit, recognizes each letter in the words