| | Cognitive function | Overview of operation |
| Centroid generation
Tasks: 1–3 | Buffers implicated: blending, imaginal, and goal
Biases instantiated: base-rate neglect, anchoring and adjustment
The model generates a category centroid by aggregating overall of the perceived events (SIGACTs) in memory via the blended memory retrieval mechanism. Judgments are based on generating a centroid-of-centroids by performing a blended retrieval over all previously generated centroids, resulting to a tendency to anchor to early judgments. Because there is an equal number of centroids per category, this mechanism explicitly neglects base rate |
| Path planning
Tasks: 3-4 | Buffers implicated: retrieval, imaginal, and goal
Biases instantiated: anchoring and adjustment
The model parses the roads into a set of intersections and road segments. The model hill-climbs by starting at the category centroid and appends contiguous road segments until the probe event is reached. Road segment lengths are perceived veridically; however, when recalled the lengths are influenced by bottom-up perceptual mechanisms (e.g., curve complexity and length) simulated by a power law with an exponent less than unity. This leads to underestimation of longer and curvier segments, resulting in a tendency to anchor when perceiving long segments |
| Probability adjustment
Tasks: 1–6 | Buffers implicated: blending, imaginal, and goal
Biases instantiated: anchoring in weighing evidence, confirmation bias
The model represents the prior probability and multiplicative factor rule and then attempts to estimate the correct posterior by performing a blended retrieval over similar chunks in memory in a form of instance-based learning. The natural tendency towards regression to the mean in blended retrievals leads to anchoring bias in higher probabilities and confirmation bias in lower probabilities. The partial matching mechanism is used to allow for matches between the prior and similar values in DM |
| Resource allocation
Tasks: 1–6 | Buffers implicated: blending, imaginal, and goal
Biases instantiated: probability matching
The model takes the probability assigned to a category and then estimates an expected outcome by performing a blended retrieval using the probability as a cue. The outcome value of the retrieved chunk is the expected outcome for the trial. Next, an additional blended retrieval is performed based on both the probability and expected outcome, whose output is the resources allocation
After feedback, the model stores the leading category probability, the resources allocated, and the actual outcome of the trial. Up to two counterfactuals are learned, representing what would have happened if a winner-take-all or pure probability matching resources allocation had occurred. Negative feedback on forced winner-take-all assignments in Tasks 1–3 leads to probability matching in Tasks 4–6 |
| Layer selection
Task: 4–6 | Buffers implicated: blending, goal
Biases instantiated: confirmation bias
In Task 6, the model uses partial matching to find chunks representing past layer-selection experiences that are similar to the current situation (the distribution of probabilities over hypotheses). If that retrieval succeeds, the model attempts to estimate the utility of each potential layer choice by performing a blended retrieval over the utilities of past layer-choice outcomes in similar situations. The layer choice that has the highest utility is selected. If the model fails to retrieve past experiences similar to the current situations, it performs a “look-ahead” search by calculating the expected utility for some feature layers. The number of moves mentally searched will not often be exhaustive
The blended retrieval mechanism will tend to average the utility of different feature layers based on prior experiences from Tasks 4 and 5 (where feature layers were provided to participants), in addition to prior trials on Task 6. |
|
|
