Fuzzy Trace Theory makes four main predictions about risky decision problems:

1. Decision options are encoded at a detailed verbatim level and a high level of abstraction – or gist – simultaneously. Effects that are not consistent with expected utility, such as framing, may arise when decision options encode different gists.

2. Choices between gist categories are made upon the basis of binary valenced affect. Decision-makers prefer the positive-affect option.

3. When categorical contrasts are not possible, decision-makers will revert to more precise gists, e.g., ordinal representations, and framing effects will be attenuated.

4. Gist categories are encoded based upon a subject's prior experience (e.g., cultural norms). These take the form of categorical contrasts, such as psychologically special representations of numbers (e.g., all, none, certainty, etc) and may be culturally contingent. Changing these will change what gists are encoded.

Rivers et al., [9] illustrate FTT with the example of an underage adolescent who must decide between two options for how to spend her evening. She can go to a fun party where alcohol is served, but which might be broken up by parents (and therefore take a risk of being culturally sanctioned); or she can go to a sleepover that is not as much fun, but is also not culturally sanctioned (and therefore not take a risk). Suppose the adolescent knows that the sleepover will be fun with certainty, whereas there is a 90% chance that the party will be twice as much fun as the sleepover, but there is a 10% chance that the parents will shut the party down, which is no fun. A classically rational decision-maker would play the odds – i.e., they would attend the risky party because, on average, the party is likely to be 1.8 times as fun as the sleepover. In contrast, a decision-maker who relies only on gist interpretations would perceive the choice as between some fun with certainty (the sleepover) and maybe some fun but maybe no fun (the party). This gist-based decision-maker would choose the sleepover. This gist hierarchy has been implemented as part of a computational model of decision-making under risk, which makes predictions for outcomes. We next describe our theory and illustrate its workings with examples from the computer program.

We begin by representing our decision problem in a decision space, where each axis corresponds to a variable of interest. Our computer program first prompts the user to indicate the number of axes in the decision space, and the names of these axes. In the adolescent decision problem, one axis would represent some quantity of fun, whereas a second axis would represent the probability with which that fun is attained. Each point in our space is a complement in a decision option.

The gist routine partitions this space into categories, each representing a set of points in the space. According to the theory, these categories may arise from the decision-maker's prior knowledge and expectations, as well as psychological regularities. For example, there is strong evidence to indicate that some of a quantity is psychologically different than none of a quantity [10-13], leading most decision-makers to distinguish between these categories. Similarly, “all” of a quantity is treated qualitatively distinct manner [14]. Analogously, no chance (probability = 0%) and certainty (probability = 100%) are psychologically distinct values on the probability axis [7].

After collecting information about the decision space (i.e., what are its axes/major variables), our program prompts the user to enter constraints on the decision space. These constraints take the form of an algebraic expression that is set equal to zero.

Found a mistake? Please highlight the word and press Shift + Enter