I think one can accept all the direct factual claims of the paper while having a fairly different interpretation. I certainly agree with the author that chickens can watch someone put food somewhere out of sight, and then have a model of the world that involves a reward if they go towards the place the food went. And that they can be trained to do a few tasks via simple reinforcement learning, and have pretty good sensory processing. This is cognitive behavior much more impressive than that of the bonsai tree, which has much more rudimentary reinforcement learning and can only be trained to do tasks on longer timescales, and has no concept of object permanence at all. I just don’t think it’s very likely that chickens have introspection, deep understanding of the environment, the ability to make novel multi-step plans, or even particularly good episodic memory.
Erm, some people would disagree with that.
Nice review article!
I think one can accept all the direct factual claims of the paper while having a fairly different interpretation. I certainly agree with the author that chickens can watch someone put food somewhere out of sight, and then have a model of the world that involves a reward if they go towards the place the food went. And that they can be trained to do a few tasks via simple reinforcement learning, and have pretty good sensory processing. This is cognitive behavior much more impressive than that of the bonsai tree, which has much more rudimentary reinforcement learning and can only be trained to do tasks on longer timescales, and has no concept of object permanence at all. I just don’t think it’s very likely that chickens have introspection, deep understanding of the environment, the ability to make novel multi-step plans, or even particularly good episodic memory.