PSYCH 240 1ST Edition Lecture 5Outline of Last Lecture: Bottom-Up ProcessingI. Pattern RecognitionII. Template TheoryIII. Feature TheoryIV. Structural DescriptionsV. Next Monday’s LectureOutline of Current Lecture I. Top-Down ProcessingII. Top-Down Bias EffectsIII. Top-Down and Word Superiority Effects in the Real WorldIV. Interactive Activation in the BrainCurrent Lecture: Top-Down ProcessingI. Top-Down ProcessingA. The same stimulus comes in from the environment but our existing knowledge, expectations, or goals bias us to perceive it one way or anotheri. See a bunny1. Caveman thinks “food”2. Little girl thinks “pet”B. Definitionsi. “Top”: refers to areas of brain responsible for higher level cognition such as: 1. Goal setting, decision-making, language, memoryii. “Bottom”: refers to lower-level areas of brain that receive input from sensation1. Examples: vision, auditioniii. Top-Down Processing: refers to processing that originates in higher cognition and proceeds downwards towards sensationiv. Bottom-Up Processing: refers to processing that originates in sensory areas and proceeds upwards towards areas responsible for higher cognitionII. Top-Down Bias EffectsThese notes represent a detailed interpretation of the professor’s lecture. GradeBuddy is best used as a supplement to your own notes, not as a substitute.A. Expectation/Bias: i. your own expectations or biases that can affect the way you perceive something1. Suppose you’re a juror – you hear a bunch of evidence and it’s your job to decide whether they are guilty beyond a reasonable doubta. High threshold – w/o a doubt2. Suppose you’re a radiologist who looks at x-rays and you’re supposed to findevidence of tumors. Now you want a relatively low threshold for detecting stimuli b/c you don’t want to miss a tumora. Low threshold - don’t want to miss anything so if there’s any chance,flag itii. Signal Detection: detect some “signal” in the presence of noise/distractions1. A frequently encountered taska. Detecting a tumor on an x-ray1.Tumor is signalb. Looking for a particular exit on the highway1.Particular exit is signal2. Signal Detection Theory (SDT) addresses how we make these detection decisions3. Demonstrates how expectations/biases (top-down influences) effect perception4. Conceptsa. Signal: something in the environment you are trying to detectb. Noise: things in environment other than the signalc. Sensitivity: how easy/difficult it is to discriminate signal from noisei. Good (high sensitivity): great at distinguishing signal from noiseii. Poor (low sensitivity): every noise you hear around you is an ax murderer who is coming to kill the camping party – you don’t differentiate btwn the signal and the noise2.Affected byi. How distinctive the distal stimuli area. the signal is a lot louder than the noiseii. Your own hearinga. Some people have good hearing, some people have bad hearingd. Bias: your tendency to say “yes” vs. “no”. Determined by expectations or payoffs 1.Effects of changing biasi. Examplea. Keep signal and noise the same but set a very high threshold (need to be absolutely sure for a “yes” responsei. Significantly lower signals are heard5. Signal Detection Matrixa. Hit: there is a signal AND you correctly detect itb. Miss: there is a signal BUT you fail to detect itc. Correct Rejection: There is no signal AND you correctly say there was no signald. False Alarm: there is not signal BUT you say that there was a signal* Accuracy: % Hits + % Correct RejectionsStimulus EventResponsePresent Not Present“Yes” Hit False Alarm“No” Miss Correct RejectionB. Signal Detection Experimentsi. Experiment #11. 200 trials (100 signal present, 100 not present)2. Payoffs bias subjects for “YES”a. You have a “hit” = $10 correct rejection = $1b. False Alarm or Miss = lose $13. Going to have a low threshold because you don’t want to miss opportunity to get $10a. Bias to produce “yes” responses 4. Result: accuracy = (90+60)/200 = 75%ii. Experiment #21. 200 trials (100 signal present, 100 not present)2. Payoffs bias subjects for “NO”a. Correct Rejection = $10 Hit = $1b. False Alarm/Miss = lose $13. Unless you’re sure you heard the signal, you’re going to say noa. Bias to produce “no” responses4. Result: accuracy – (60+90)/200 = 75%iii. Important1. Bias: tendency to say “yes” vs “no”a. It moves your responses btwn the “yes” and “no” rows of the matrix2. Sensitivity: how easy it is to discriminate the signal from the noisea. Depends on how separable are signal and noiseb. Depends on how good your “detector” is (vision, hearing, sense of touch)3. Accuracy depends on: a. Proportion of trials w/ signal present/not presentb. Biasc. SensitivityC. Contexti. Effect of Context on Perception1. Signal/detection: example of how biases/expectation affect perception2. Context effects: when perception of an object is affected by its context/environmenta. Examples1.Subjective contoursi. We perceive triangles even though they’re not there (picture: 3 angle brackets and 3 pac-man shapes)ii. Can almost see an edge of the white triangle that simply isn’t therea. Can find evidence of visual cortical cells firing for an edge at that spot, even though it doesn’t exist2.Letter recognitioni. What you perceive depends on contexta. In a row of numbers, you perceive the signal as 13 but in a row of letters you perceive it as Bb. Objects out of context1.Experiment: subjects asked to identity objects in picturesi. Given image where fire hydrant is on top of a mailboxa. Took way longer to identify objectii. Shown another image where fire hydrant is on the sidewalk2.Results: performance was significantly worse when objects out of contextD. Higher Levels of Analysis that Affect Lower Onesi. Word Superiority Effect (Reicher, 1969): people are better at recognizing letters in the context of words than alonea. Q: is it easier to identify letters when they:1.Appear as part of a word?2.Appear alone?b. ANSWER: When they appear as part of a word1.Subjects better at word condition than letter condition2.Supports existence of “word superiority effect”1. Reicher’s Initial Experimenta. Presented words or letters by themselves. Flashed briefly then gave an arrow indicating what he wanted subject to identify. Such as, what was the last letter of that word or what was that letter?i. Results showed that people are better at