TYPES OF HEURISTICS: Fast & Frugal Heuristics vs. Other Methods
selected from Gigerenzer & Todd Simple Heuristics That Make Us Smart Oxford 1999

(compiled by Edward G. Rozycki, Ed.D. )

edited 10/18/18


other methods*


Step 0

Step 1: Search Rule

Step 2: Stopping Rule

Step 3: Decision Rule

Notes w/ pg #'s

Recognition RH Chap 2

Pairwise random presentation of objects (PRPO)

See if item of selected pair is recognized.

If neither or both recognized, back to 0.

If only one and not more than one item recognized, stop

Select recognized object.

Heuristic def's:

ecological validity:

discrimination rate:

Conflicts avoided 81

Rules non-compensatory 81

Minimalist MH 79


Cue randomly selected. Use RH

Random selection of cue; if none found, guess.

If cue values are 1,0 go to 3; otherwise step 1

Object with 1 has higher value.


Take the Last TLH 80

History of Cue used to solve previous problem: Einstellung**;)

(PRPO) Cue randomly selected. Use RH.

Einstellung** From second problem, start with last problem's successful cue.

If not stopped, use cue antecedent to failed one from previous problem.

Object with 1 has higher value. Keep history of successful cues.


Take the Best TBH 80

Order by perceived validities


(PRPO) Cue NOT randomly selected. Use RH.

Ordered Search: Use cue of highest validity not yet tried for present task..

If choice fails, used next highest validity cue.

Compare cue values Object with 1 has higher value.


Franklin 76 (adapted) weightings not subjective as were Franklin's

Weighted linear combination of cues (WLCC)



Weightings may vary.

When all cues and weights determined

CMI by sum of products of weights by cues. Select highest product.

"Commandments" of unbounded rational judgment:

a. complete search (CS): find all information available

b. compensation (CAI): combine all pieces of information



Linear strategy, unit weights



Weights all = 1. Determine all cues.

See Franklin Step3.


Multiple Regression

Attempt to minimize least squares in regression line thru all data points



Expected Value of CMI

Accept regression with least sum of squares.


Expected Value



All values with associated probabilities


Choose highest of all E(V)'s.


See also Gigerenzer & Todd, p. 143

**Einstellung: attitude, setting, stand