04.+Expectations

February 17, 2015
 * 04. Expectations**


 * Top Down & Bottom up**

Example: tapping the Itsy-Bitsy Spider
 * first time you just hear tapping
 * second time you hear the song

Example: Hidden messages in songs played backwards


 * Bottom-up processing:**
 * Data driven
 * Incoming information influences perceptions
 * you hear tapping
 * Top-down processing:**
 * Concept-driven
 * Knowledge and expectations influence perceptions
 * you hear a song

For processes such as reading:
 * Object Recognition**

Palmer (1975): participants viewed a scene (e.g., a kitchen) for 2 seconds. Objects were flashed briefly within the scene.
 * Bottom-up processing
 * Perception triggers feature detectors
 * They feed upward.
 * Top-down processing[[image:B or 13?.png width="229" height="158" align="right"]]
 * Knowledge and expectations trigger complex pattern detectors (e.g., letters)
 * They feed downward
 * Results (identification accuracy):
 * Context appropriate object (bread): 80%
 * Context inappropriate object (mailbox): 40%
 * The kitchen created an expectation; expectations affect perception.

Lee, Frederick, & Ariely (2006): Try two beers. One is MIT Brew and one is familiar.
 * Food Preferences**
 * 3 conditions:
 * Blind: sample beers, indicate preference
 * Before: told the secret ingredient for the MIT Brew is vinegar (gross), then sample beers, then indicate preference
 * After: sample beers first, then told it's vinegar, then indicate preference
 * Result: Before condition less likely to prefer MIT brew

Other examples:
 * Coke is rated as tastier when it's served in a brand-name cup (McClure et al., 2004)
 * Yogurt and cheese spreads rated as yuckier if they are labeled "low-fat" (Wardle & Solomons, 1994)
 * Wine labeled as more expensive results in pleasure-related brain areas (Plassman et al., 2008)

Stillman (1993): Participants were given one drink. They were told "the color is independent of the taste." Actual flavor: orange or raspberry. Colors: clear, red, yellow, green.
 * Expectations affect flavor identification

Crum et al. (2011): Participants consumed milkshakes on two different days. Both were 380 calories. Before consuming, they read the labels:
 * "Indulgence": 620 calories, looks unhealthy and delicious and fatty...
 * "Sensi-Shake": 140 calories, guilt-free, good for you...
 * Ask how healthy it tastes: Sensi-shake perceived as more healthy
 * Ghrelin is a gut peptide that is secreted when your stomach is empty. Basically, ghrelin makes you feel hungry.
 * Measured ghrelin levels before looking at label, before drinking, and after drinking
 * Ghrelin measurement results (more ghrelin = more hungry)
 * No info, ghrelin levels are the same
 * See label: More ghrelin in indulgent condition than in sensi-shake condition
 * After drinking: More ghrelin in sensi-shake condition than indulgent condition
 * Even when actually calories were the same, expectations affect satiety.

Durgin et al. (2009): Asked participants to judge: "How steep is that slope?" Schnall et al., (2010): Participants don't eat for 3 hours. Then drink something with sugar, or no sugar. They wait 10 minutes, put on a backpack, and estimate the slope HOWEVER, these interpretations may not be correct:
 * Steepness**
 * Backpack condition + no backpack condition
 * Result: A heavy backpack appears to make the slope look steeper
 * Low blood sugar appears to make the slope look steeper.
 * Durgin et al. (2009): 3 conditions: no backpack, backpack, and backpack + cover story (told their ankle muscles were being monitored. used real electrodes, etc.). Asked how steep the slope is.
 * Participants in the backpack condition were asked why they had a backpack. 10/11 participants guessed the backpack was supposed to affect their judgments.
 * Result: No backpack and backpack + cover story conditions were the same. Backpack condition judged the slope to be more steep.
 * Therefore: The heavy backpack did not make the slope seem steeper.

Potchen (2005): radiologists viewed x-rays and asked to classify it as "normal" or "abnormal"
 * Medicine**
 * Conditions:
 * No cue (control)
 * "This x-ray is from a routine physical"
 * "This x-ray is from a survey for cancer"
 * Results: Cancer survey condition had higher percentage of correct diagnoses

Hatala et al. (1999): Medical residents made diagnoses based on EKGs and irrelevant demographic information. Residents first reviewed a series of cases consisting of a demographics and an EKG, and then worked on a series of test cases. The EKGs should be unequivocal.
 * 2 conditions:
 * Matched Demographics Group: demographic info from the review cases was matched to the test cases
 * Mismatched Demographics Group: demographics from the review cases were not matched with the test cases
 * Result: Matched demographics had a lower percentage of correct diagnosis than mismatched demographics. Test cases were always a different diagnosis, so if subjects recalled the review case, they were less likely to get the diagnosis. Irrelevant specific information influenced diagnosis



Bottom Line: Two factors have a big impact on perception
 * Bottom-up processing is based on data in the world
 * Top-down processing is based on expectations