Evolution and Vision

William F. Broderick

2017-04-17

Overview

Why is rodent visual system so different from human and what does that mean for vision?

Eyes
Of mice and men
Why maps?

my interest in this got started when we were taught that rodents have neither foveas nor orderly orientation columns in V1. rodents can still do orientation discrimination 2AFC-type tasks, so then what's the point of those maps?

and why not have foveation? To me, it makes sense to have fovea because of limitation: can only get so many axons through the optic disk, we take super high resolution in one point instead of kind-of-high everywhere.

before I get started, does anyone have any ideas?

This will kind of be about evolution, but more about ethology, comparing visual systems of different animals. Unfortunately, most of what we know is about structure of eyes (and so the optics), not on the functional properties. I think relatively few species have had their V1 characterized.

I also don't think I have a single equation in this presentation, so that's new for us

things we'll talk about: (make all appear)

What's the point of vision?

"If the question is how the brain leads to behavior, we are first asking why is the brain performing this behavior and then asking how is it doing it" – Krakauer et al, 2017

Light-controlled behaviors

Behaviors controlled by ambient light (circadian rhythms)
Behaviors based on directional light (phototaxis)
Tasks requiring low spatial resolution (object avoidance, orientation to sun)
Tasks requiring high spatial resolution (detection and pursuit of prey, recognition of individuals)

Light-sensitive organs

Land2010-Fig1.5.jpg

Parallel evolutions of eyes

Land2010-Fig1.9.jpg

Eye properties

Resolution: how finely the optical environment is sampled
Sensitivity: how many photos your receptors receive

To make either one better: increase eye size

OWLS

Pupils and light levels

Land2010-Fig5.11.jpg

Binocular vision

Land2010-Fig5.13.jpg

Ganglion cell distribution

Land2010-Fig5.12.jpg

Hawk acuity

How similar is mouse vision to primate vision?

Ventral visual stream?

Receptive field size

Decoding object identity from mice

Decoding object identity from macaques

Mouse V1 orientation preferences inherited from LGN

Cross-species differences

WHY MAPS?

Minimizes wiring cost

why maps? anatomical reason and possibly functional ones, what do people think?

Chklovskii2004-maps-in

maps probably come minimizing wiring cost
basically, function inherited from connectivity, and connecitivty + wiring constrain gives rise to maps
with a wiring cost penalty, you get different maps based on how cells connect to each other
doesn't answer question of why you'd have different functional layouts in different animals or why one would be preferred over the other (retinotopy easy to argue for)

Nauhaus2014-build-maps:

well-arranged maps make sampling different portions of the stimulus space much easier and, if you have two orthogonal ones at the same time, it's easy to get all possible permutations of the properties
with maps, sampling different portions of the it will yield different response distributions, because there's structure and not "white noise"