Marr makes a distinction between the visual systems of flies and other animals and of humans, stating that the former are more task-specific, and the latter is more general. He gives an evolutionary explanation that relates specific algorithms of fly vision to specific survival purposes in the real world in flight control. The notion that human vision is much more general makes a lot of sense, given that human cognition is much more general overall. That said, Marr concedes that there are similarly special-purpose mechanisms in human vision – he gives the examples of directing the eyes to a sudden unexpected movement, and blinking or dodging when something quickly approaches one's head. But more complex mechanisms of human vision, such as a predisposition for facial recognition, also have domain-specific, evolutionary aspects. Although these are more computationally complex than the fly's flight-control algorithms, there are still genetic constraints involved. Relatedly, in terms of the "grandmother cell" idea, there are findings in cognitive neuropsychology that show that certain distinct neural patterns do consistently light up in response to specific entities (I recall a study implying "Jennifer Aniston neurons"...). Taken these together, my question is, even if human vision and broader perception are much more general purpose and complex, where are the limits of this generality due to genetic, task-specific constraints similar to those in animals?