Utility-weighted sampling in decisions from experience

Published on 2015-07-282125 Views

People overweight extreme events in decision-making and overestimate their frequency. Previous theoretical work has shown that this apparently irrational bias could result from utility-weighted sampl

RLDM 2015 - Edmonton

Presentation

Utility-weighted sampling in decisions from experience00:00

Extreme potential outcomes - 102:58:58

Extreme potential outcomes - 208:32:07

Extreme potential outcomes - 315:44:35

Extreme potential outcomes - 420:44:49

Extreme potential outcomes - 522:14:24

Expected Utility Theory - 123:01:27

Expected Utility Theory - 226:16:28

Expected Utility Theory - 327:45:05

Expected Utility Theory - 434:16:09

Expected Utility Theory - 538:41:04

EU can be Approximated by Sampling - 142:47:14

EU can be Approximated by Sampling - 245:40:14

EU can be Approximated by Sampling - 346:55:18

EU can be Approximated by Sampling - 449:25:05

EU can be Approximated by Sampling - 555:44:39

Representative sampling is dangerous - 161:32:42

Representative sampling is dangerous - 267:10:08

Representative sampling is dangerous - 374:24:55

Representative sampling is dangerous - 478:34:31

Representative sampling is dangerous - 579:34:07

Representative sampling is dangerous - 690:52:22

Utility estimation by importance sampling - 196:46:31

Utility estimation by importance sampling - 2101:51:02

Utility estimation by importance sampling - 3104:27:22

Utility estimation by importance sampling - 4109:04:51

Utility estimation by importance sampling - 5112:06:39

Utility estimation by importance sampling - 6116:34:38

Utility estimation by importance sampling - 7117:04:34

Utility estimation by importance sampling - 8125:27:03

Answer: Utility-Weighted Sampling (UWS) - 1127:15:10

Answer: Utility-Weighted Sampling (UWS) - 2133:03:41

Decisions from Experience (Ludvig, et al., 2014) - 1137:59:19

Decisions from Experience (Ludvig, et al., 2014) - 2147:36:57

Decisions from Experience (Ludvig, et al., 2014) - 3149:05:55

Decisions from Experience (Ludvig, et al., 2014) - 4150:55:32

Decisions from Experience (Ludvig, et al., 2014) - 5152:00:09

Decisions from Experience (Ludvig, et al., 2014) - 6152:45:14

Inconsistent Risk Preferences Emerge from Learning154:55:16

UWS Can Emerge from Reward-Modulated Associative Learning - 1166:21:04

UWS Can Emerge from Reward-Modulated Associative Learning - 2169:20:31

UWS Can Emerge from Reward-Modulated Associative Learning - 3170:27:51

UWS Can Emerge from Reward-Modulated Associative Learning - 4171:17:16

UWS Can Emerge from Reward-Modulated Associative Learning - 5175:56:31

UWS Can Emerge from Reward-Modulated Associative Learning - 6179:16:29

UWS Can Emerge from Reward-Modulated Associative Learning - 7187:17:12

UWS Can Emerge from Reward-Modulated Associative Learning - 8187:28:49

Learning Rule Convergences to Utility-Weighted Sampling - 1191:23:49

Learning Rule Convergences to Utility-Weighted Sampling - 2196:19:35

Efficient coding (Summerfield & Tsetsos, 2015) - 1199:33:08

Efficient coding (Summerfield & Tsetsos, 2015) - 2202:33:23

Efficient coding (Summerfield & Tsetsos, 2015) - 3204:51:09

Model fitting210:49:02

UWS captures that people learn to overweight extreme outcomes - 1216:18:43

UWS captures that people learn to overweight extreme outcomes - 2218:52:52

Utility-Weighted Sampling Captures Memory Biases (Madan et al. 2014) - 1232:17:43

Utility-Weighted Sampling Captures Memory Biases (Madan et al. 2014)- 2238:36:17

Utility-Weighted Sampling Captures Memory Biases (Madan et al. 2014) - 3240:32:08

Utility-Weighted Sampling Captures Memory Biases (Madan et al. 2014) - 4242:02:15

Utility-Weighted Sampling Captures Memory Biases (Madan et al. 2014) - 5243:46:18

Utility-Weighted Sampling Captures Memory Biases (Madan et al. 2014) - 6243:58:44

Biased Beliefs Predict Risk Seeking - 1244:50:25

Biased Beliefs Predict Risk Seeking - 2247:34:28

Biased Beliefs Predict Risk Seeking - 3247:51:07

Biased Beliefs Predict Risk Seeking - 4248:06:24

Conclusions - 1250:22:51

Conclusions - 2251:18:37

Conclusions - 3254:04:03

Conclusions - 4256:19:52

Conclusions - 5262:22:01

Thank You265:53:13