Optimal Bayesian models have been highly successful in describing human performance on perceptual decision-making tasks, such as cue combination and visual search. However, recent studies have argued ...that these models are often overly flexible and therefore lack explanatory power. Moreover, there are indications that neural computation is inherently imprecise, which makes it implausible that humans would perform optimally on any non-trivial task. Here, we reconsider human performance on a visual-search task by using an approach that constrains model flexibility and tests for computational imperfections. Subjects performed a target detection task in which targets and distractors were tilted ellipses with orientations drawn from Gaussian distributions with different means. We varied the amount of overlap between these distributions to create multiple levels of external uncertainty. We also varied the level of sensory noise, by testing subjects under both short and unlimited display times. On average, empirical performance-measured as d'-fell 18.1% short of optimal performance. We found no evidence that the magnitude of this suboptimality was affected by the level of internal or external uncertainty. The data were well accounted for by a Bayesian model with imperfections in its computations. This "imperfect Bayesian" model convincingly outperformed the "flawless Bayesian" model as well as all ten heuristic models that we tested. These results suggest that perception is founded on Bayesian principles, but with suboptimalities in the implementation of these principles. The view of perception as imperfect Bayesian inference can provide a middle ground between traditional Bayesian and anti-Bayesian views.
Base rate neglect refers to people's apparent tendency to underweight or even ignore base rate information when estimating posterior probabilities for events, such as the probability that a person ...with a positive cancer-test outcome actually does have cancer. While often replicated, almost all evidence for the phenomenon comes from studies that used problems with extremely low base rates, high hit rates, and low false alarm rates. It is currently unclear whether the effect generalizes to reasoning problems outside this “corner” of the entire problem space. Another limitation of previous studies is that they have focused on describing empirical patterns of the effect at the group level and not so much on the underlying strategies and individual differences. Here, we address these two limitations by testing participants on a broader problem space and modeling their responses at a single-participant level. We find that the empirical patterns that have served as evidence for base-rate neglect generalize to a larger problem space, albeit with large individual differences in the extent with which participants “neglect” base rates. In particular, we find a bi-modal distribution consisting of one group of participants who almost entirely ignore the base rate and another group who almost entirely account for it. This heterogeneity is reflected in the cognitive modeling results: participants in the former group were best captured by a linear-additive model, while participants in the latter group were best captured by a Bayesian model. We find little evidence for heuristic models. Altogether, these results suggest that the effect known as “base-rate neglect” generalizes to a large set of reasoning problems, but varies largely across participants and may need a reinterpretation in terms of the underlying cognitive mechanisms.
Studies in perception have found that humans often behave in accordance with Bayesian principles, while studies in higher-level cognition tend to find the opposite. A key methodological difference is ...that perceptual studies typically focus on whether people weight sensory cues according to their precision (determined by sensory noise levels), while studies with cognitive tasks concentrate on explicit inverse inference from likelihoods to posteriors. Here, we investigate if lay-people spontaneously engage in precision weighting in three cognitive inference tasks that require combining prior information with new data. We peel the layers of the "intuitive Bayesian" by categorizing participants into four categories: (a) No appreciation for the need to consider both prior and data; (b) Consideration of both prior and data; (c) Appreciation of the need to weight the prior and data according to their precision; (d) Ability to explicitly distinguish the inverse probabilities and perform inferences from description (rather than experience). The results suggest that with a lenient coding criterion, 58% of the participants appreciated the need to consider both the prior and data, 25% appreciated the need to weight them with their precision, but only 12% correctly solved the tasks that required understanding of inverse probabilities. Hence, while many participants weigh the data against priors, as in perceptual studies, they seem to have difficulty with "unpacking" symbols into their real-world extensions, like frequencies and sample sizes, and understanding inverse probability. Regardless of other task differences, people thus have larger difficulty with aspects of Bayesian performance typically probed in "cognitive studies."
Elucidation of the key role of thrombosis in cardiovascular disease events has arisen considerable interest in hemostatic factors and in the repeatability of their determinations. Data on long-term ...repeatability has, however, remained scanty. We examined twice 208 men and 265 women in North Karelia, eastern Finland. The baseline examination was a part of the FINRISK 1992 Hemostasis Study and the age-range of participants was between 45-64 years. The re-examination took place three years later in 1995. Both surveys followed the same protocol and were carried out during the same season. Spearman rank correlation coefficients between 1992 and 1995 measurements of fibrinogen, factor VII coagulant activity (FVII:C), factor VII antigen (FVII:Ag), and plasminogen were among men 0.72, 0.77, 0.46 and 0.56, respectively. For total cholesterol, HDL-cholesterol, triglycerides and diastolic blood pressure the corresponding coefficients were 0.74, 0.83, 0.66, and 0.54. In women, the coefficient of fibrinogen was lower than in men, 0.62, otherwise the results were similar. Of men belonging to the highest quarter of fibrinogen, FVII:C, FVII:Ag and plasminogen in 1992, 65%, 60%, 53% and 60% belonged to the highest quarter of respective distributions also in 1995. In women, the corresponding proportions were 64%, 65%, 46% and 58%. The modest repeatability of FVII:Ag and plasminogen was mainly due to the high intraindividual variability. However, in comparisons of plasma levels between two groups, relatively small sample sizes seemed to give adequate statistical power to detect possible differences in FVII:Ag and plasminogen. In conclusion, the long-term repeatability of fibrinogen and FVII:C is similar to that of triglycerides and even better than that of diastolic blood pressure, but somewhat lower than the repeatability of total cholesterol. FVII:Ag and plasminogen did not have very good repeatability and more than one measurement of them should be considered if they are used as predictors of cardiovascular disease in prospective studies.