Very happy to announce that my new paper with Brian Scholl, "Retinotopic adaptation reveals distinct categories of causal perception", has just been published in Cognition!

Free here for 50 days: https://authors.elsevier.com/a/1bRfA_EbvtkBE 

I'm super excited about this paper. A brief summary:
1/15

We don't just make judgments about cause and effect, we truly perceive it. For example, Michotte's launching event (see movie) is truly *perceived* as causal, regardless of what you *think* about it. 2/15

One of the strongest (and just straight-up most awesome) pieces of evidence for causal *perception* is retinotopically-specific visual adaptation to causality, first reported by @MartinRolfs and colleagues in 2013.

Let me unpack what that actually means. 3/15

"Visual adaptation to causality" = if you watch hundreds of launching events, subsequent ambiguous events that can look causal or non-causal will tend to look non-causal. It's kind of analogous (but not actually related) to the waterfall illusion ( https://en.wikipedia.org/wiki/Motion_aftereffect)

4/15

"Retinotopically-specific" = The adaptation effect ONLY works when the adaptation stream and test events are presented to the same location *on the retina*.

Launching is processed early enough that the visual system still uses a map of the retina as its frame of reference! 5/15

We used this adaptation effect to identify natural "joints" in causal perception.

To do this, we used adaptation transfer: If you adapt to one event and test on another and the retinotopically-specific adaptation effect still works, they must share the adapted feature. 6/15

So is this adaptation to "causality" writ large, or to "launching" very narrowly? To test this, we adapted participants to two other causal events.

The first was "triggering", which is identical to launching except B moves 3x faster than A (see movie) 7/15

The other was "entraining", in which A does not stop and continues moving with B (see movie).

Triggering violates a Newtonian constraint on elastic collisions, while entraining is a constant application of force, not a simple elastic collision.

What did we find? 8/15

Adaptation to triggering affects the perception of launching.

Adaptation to entraining does *not*.

The visual system has distinct types of causal perception, but some events (notably launching and triggering) share an underlying causal representation.

9/15

This pattern is something of a surprise, given my 2017 Psych Science paper showing that triggering is an "oddball" in a visual search array of launching events, and that infants habituated to launching dishabituate to triggering!

But these methods capture different things. 10/15

This paper says that launching and triggering are the same category of causal event, and my 2017 paper says that they are still meaningfully different in another way. My view: We are sensitive to the physics of this category of event, but we need to ID the category first.*

11/15

*Brian doesn't fully endorse this specific account, but the general logic is that the visual system can make distinctions at multiple levels. Motion aftereffects work across waterfalls and falling sand, even though we can see they are different things. 12/15

This paper gives the strongest evidence to date that there truly are *multiple* causal perceptions in the mind, where 75+ years of research had assumed there was only one.

It also provides a powerful new tool for characterizing the structure of causal perception. 13/15

I'm very proud of this project. Visual adaptation to causality (or some types of causality) is just an awesome phenomenon, and I loved being able to replicate it and show how robust it is (three times in this paper alone!).

14/15

It also opens up so many exciting questions that I'm still just starting to explore. First and foremost: "what exactly *is* entraining, then?"

I have some ideas, but that's another paper...which I should get back to writing.

Happy to answer questions! /end