smh the bots are just repeating themselves at this point

One might worry about low-level confounds, e.g., with the colors that we chose. We think those are unlikely to explain the effects we observe. But for good measure, we showed that same "value" adaptation with symbolic numbers. The combination of these results is hard to dismiss.

Interestingly, though, it wasn't that people adapted to everything that we put in front of them. For instance, people did not exhibit any "reverse" adaptation. We also showed that they adapted to cumulative but not average value (regardless of what they were instructed to do).

We conducted a standard adaptation paradigm, except that instead of having observers to adapt to something like number, or size, or color, or motion, we had them adapt to displays of fake "coins" with arbitrarily assigned value. Surprisingly, people exhibit "adaptation" all the same.

Neither children nor adults always match objects based on topology, though. But here's the cool part: Children's and adults' judgments are highly correlated. This suggests to us that there are essential spatial "building blocks" guiding these judgments that are stable across development.

Second, we show that children, like adults, match objects based on topology. In the example here, for instance, the left object is more similar to the sample in that the left horizontal line moved less. Yet both children and adults say that the right object is more similar to the top one.

First things first: We show that children notice *very* subtle differences in topological structure. We show them six shapes and ask them to identify which one is not like the others. No other instructions, no hints. Even though this task is hard (even for adults!), four-year-old children can do it.

In perhaps the wackiest experiment in the paper, we showed that objects that were combined in ways that were combined in ways that altered the topological structure were harder to identify subsequently (as opposed to combinations that did not alter the topological structure).

My personal favorite finding: We show that these same topological difference influence how people enumerate the number of line segments of objects -- almost as if people perceive the "L-junctions" as a single, bendable segment rather than separate line segments.

You can see the influence of topology for yourself in the following visual search example. The target object is *much* easier to find in the right display, even though the critical line segmented is shifted by the same amount in both displays.

There's several neat demonstrations of the power of subtle changes in topology. For instance, people are better at quickly identifying deviant object if there is a topological change (as opposed to a larger change that does not affect topology).

As in some of our recent work, we are interested in *network topology*, of the sort that people encounter when interacting with transit maps.

I am recruiting graduate students! I’ll review applications through both the Cognitive & Developmental areas. Interested students are encouraged to be on touch; I’d love to chat with you!

Okay I know that bluesky isn't a real thing yet, but I'm going to pretend that it is.

In the final poster session at CogSci, I'll be talking about my recent work with Liz Brannon and Sam Clarke on number adaptation. Come say hi!