 is we had a really big cohort of animals. We had 27 mice, and they did over half a million decisions. And so having this really big group of animals just presented a very different picture of how decision-making works compared to just having one or two animals. So the mice see a sequence of flashes, and they have to make a judgment about whether they're high rate or low rate. And if they're low rate, they go to the left and poke in a little reward port on the left. And if it's high rate, they go to the right and poke in a little reward port on the right. And when we first developed the task, we were really interested in how they might estimate the rate of the flashes or the count of the flashes. So all of our analyses were based on the assumption that that's what they were going to pay attention to, and that's the feature that they were going to use to make their decision. It turns out they actually had something else in mind, which is that although they were indeed influenced by the rate or the count of the stimulus, they were also influenced by an additional parameter that we hadn't really thought of, which is the overall brightness of the stimulus throughout the 1,000 millisecond duration. It was a total surprise because we didn't even think about the brightness of the flashes. We sort of assumed that it was meaningless. But sort of what we realized is that the animals don't know what we've picked as the relevant dimension. They're just sort of looking at this complicated stimulus and figuring out what was I rewarded for. And an observer who does that would naturally discover both these dimensions. I think there are two major lessons that we've learned, and I think one is we shouldn't imagine that we know what the animal strategy is until we've really probed it and tested our hypothesis about that strategy specifically and to do that for every animal because they can really be quite different in how they approach a problem. And that's kind of what allowed us also to make this comparison to observations that have been made in other species. So you would have thought that mice and rats would be similar and non-human primates and humans would be similar. But for one of the measures that we looked at, which is how sort of their strategy for combining pieces of evidence over time, the mice were like the monkeys and the rats were a little more like the humans. And so I think it points to the idea that even though two animals might be more closer to each other from an evolutionary point of view, that doesn't necessarily mean that they adopt the same strategy. If they're different animals, they're still adopted for different ethylogical niches. And in some approaches to higher cognitive problems, we might be more similar to quite different animals than we are to animals that we think of as being more like us.