Table 1 Distinction between types of tests and models, with a focus on their purpose and epistemic work

Tests

Screening tests

Allows limited comprehension of an isolated aspect of a complex biological mechanism to be mapped

Limited predictive usefulness (e.g., predicting desired drug activity)

Low fidelity (underlying mechanism does not need to be similar)

Low discriminating ability (not necessarily sensitive to, e.g., triggering factors)

Not necessarily hypothesis-driven

Low construct validity; moderate predictive validity (pharmacological isomorphism only)

Biobehavioral assays

Allows broader comprehension of a mechanism, without necessary causal analogy

Moderate predictive usefulness [e.g., studying neural bases of behavioral (dys)functions]

Low fidelity (mechanism similar, but not causally analogous)

High discriminating ability (sensitive to disturbances by definition)

High predictive validity, at best moderate construct validity

Models

Simulations

Can allow inferences and extrapolation to the human disorder, with high probability that the hypothesis thus generated is true

High fidelity (similar mechanisms with probable causal analogy)

High discriminating ability (sensitive to disturbances by definition)

High face, predictive, and construct validity (considers the need to address theoretical constructs on the etiology, symptomatology, and treatment)