Philosophy & Theory

The generality of psychological theories is sharply divided by the basic-applied distinction. While basic research aims to uncover universal invariants of human thought and behaviour independent of the current moment (How does memory work? How do people solve problems?), applied research aims to provide explanations to how people solve more proximal problems (How well do eyewitnesses identify suspects in a line-up? What are the best methods to help students learn in the modern-day classroom?). I will argue that much of what is currently considered basic research is in fact applied. To make this point, I will draw on evolutionary epistemology – the idea that evolutionary logic underlies every knowledge-generation process, including biological and cultural evolution, and human creativity. Under this view, the invariant processes that basic research should aim to explain are those that increase and decrease variance and effect change over time (like the mutation and natural selection processes of biological evolution). Basic psychological research today, on the other hand, typically focuses on what people do right now and, as such, provides only "snapshot" explanations – e.g., cataloging the current strategies for memorising things or solving math or logic problems. I’ll outline a more promising way towards universal psychological theories.

This is an in-person presentation on July 21, 2024 (10:00 ~ 10:20 CEST).

There are a few purported reasons for why formal models improve the quality of the theories they implement. For example, building a model can make both the assumptions and the implications/predictions of a theory explicit and clear. While many such features are necessary, they are not sufficient for theoretical progress. I'll argue that building and testing formal models leads to better theories because the process of doing so can help hold the theory accountable to what has been observed (data) and to what else is known (other theories). Interestingly enough, many of the norms in the mathematical psychology community work in support of this goal, and the point of this talk is to make it explicit why they are important.

This is an in-person presentation on July 21, 2024 (10:20 ~ 10:40 CEST).

Due to the ongoing replication crises in Psychology, it has bees been suggested that psychologists should make more wide-spread use of formal cognitive modelling. However, the large number of decisions researchers need to make during cognitive modelling raises doubts about whether results based on cognitive modelling will be easier to replicate. Here, we present a replication attempt of Adler and Ma’s finding that heuristic models outperform Bayesian models in a orientation discrimination task with simultaneous confidence judgments (PLoS Comp Biol, 14(11), e1006572), suggesting that human perceptual decisions are not Bayes-optimal. Our analysis, using the authors' original data but reprogramming the modelling analysis from scratch, replicated Adler and Ma’s main finding that heuristic models provide better fits to the data than the Bayesian models. However, our versions of the modelling analysis produced worse models fits than the results reported by Adler and Ma. In general, it seems rather difficult to replicate complex computational cognitive models without comprehensive details about the involved computations and the underlying computer code.

This is an in-person presentation on July 21, 2024 (10:40 ~ 11:00 CEST).