Has there been any research on how mammals predict the speed of moving objects? In particular, how do they integrate top-down information? For instance, do they have greater difficulty estimating the speed of a fast-moving sloth more than a moving mouse?
Two areas in the visual cortex are associated with the perception of velocity, namely:
Both V1 and MT contain neurons that respond strongly to motion in a particular direction and also have selective responses to a particular speed. Direction and speed together define the velocity vector, i.e., these V1 and MT neurons are velocity-tuned. The velocity tuned neurons in V1 typically project to velocity tuned cells in MT. While V1 contains approximately 25% velocity-tuned neurons, those in the MT neurons are nearly all velocity-tuned (Bradley & Goyal, 2008).
Now this is all bottom-up processing. You basically ask as well whether top-down issues play a role. I do not think so. The visual system is basically split in a dorsal 'what' and a ventral 'where' stream (Fig. 1). The 'what' system identifies a target and the 'where' system (containing MT) determines location and velocity. As far as I know, there is no feedback from the dorsal stream onto the ventral one to affect velocity processing.
Fig. 1. Dorsal (parietal) and ventral (temporal) streams. source: Lehky & Sereno (2007)
@AliceD gave a great answer on the bottom up pathway. Let me add some top-down processing bits into the picture, since that's what you are asking.
First of all, at a perceptual level, motion perception is biased by prior expectations and strength of motion signal. E.g., see:
- Stocker, A. A. and Simoncelli, E. P. (2006). Noise characteristics and prior expectations in human visual speed perception. Nature Neuroscience, 9(4):578-585. http://dx.doi.org/10.1038/nn1669
The neural activity of MT (and behavior) is known to be modulated by attention. E.g., see:
- Monkey Area MT Latencies to Speed Changes Depend on Attention and Correlate with Behavioral Reaction Times. Neuron. 78(4):740–750. http://doi.org/10.1016/j.neuron.2013.03.014