Color perception in the intermediate periphery of the visual field Thorsten Hansen, Lars Pracejus & Karl R. Gegenfurtner Abteilung Allgemeine Psychologie.

Color perception in the intermediate periphery of

the visual field

Thorsten Hansen, Lars Pracejus & Karl R. Gegenfurtner

Abteilung Allgemeine Psychologie

Introduction Perception in the periphery

Contrast sensitivity

Fine details (high spatial frequencies) are harder to detect in the periphery.

This can be compensated by magnifying the image.

Introduction

Cone density in the human retina Curcio et al. (1990)

Steep decline towards periphery.

Nasal retina has a higher density than temporal retina.

Nasal Fovea Temporal

1000 /mm²

Introduction Color opponency in the fovea

Derrington (2001).

The RF center of a parvo ganglion cell in the fovea is driven by a single cone.

Introduction Color opponency at larger eccentricities

…is expected to be weaker and will eventually be zero when the center and surround is driven by an equal ratio of L:M cells

Fovea Periphery

Introduction Psychophysics vs. physiology

Psy

chop

hysi

cs (

Hum

an)

Cel

l rec

ordi

ng (

Mac

aque

)

Macaque ganglion cells remain color sensitive, while human performance drops.

Martin et al. (2001). Nature.

Introduction Psychophysics

Mullen, Sakurai & Chu (2005). Perception.

»Thus we conclude that there is little or no L/M cone opponent response measurable psychophysically beyond 20–30 deg of

eccentricity in the nasal visual field.«

LumS−(L+M)

L−M

Methods Setup: Elumens Vision station

Methods Cone-opponent axes

Methods

DKL color spaceAn achromatic axis: Lum

Two chromatic axes: L−M and S−(L+M))

L−M

S−(L+M)Lum

L−MS−(L+M)

Derrington Krauskopf Lennie

Short presentation stimuli (500 ms)

forced choice (4AFC)

threshold measured by standard stair-case procedure

Exp 1 & 2: Detection & Identification Exp. 3: Discrimination

Methods Procedure

Results Chromatic detection (5 deg

stimulus)

N = 7

in %

10°20°

30°40°

50°

Results Identification (5 deg stimulus)

in %

10°

50°

N = 7

Results Chromatic detection (8 deg

stimulus)

N = 3

L−M

S−(L+M)

Eccentricity (deg)

Lum

Comparison color varied along 8 chromatic directions

Methods Chromatic discrimination

Ellipse was fit to the data to characterize discrimination performance

Results Control: Foveal discrimination

…as expected:

Best at the adaptation point

Elongated along the saturation axesKrauskopf & Gegenfurtner (1992). Vision Res.; Hansen, Giesel & Gegenfurtner (in press), J. Vision.

Results Discrimination at 50°

Larger size of ellipses: Discrimination is worse, but not absent!

Greatest increase along L−M axis

Leads to rounder shapes of ellipses off the L−M axis

Summary Chromatic processing in the periphery

• Detection

• Identification

• Discrimination

As long as the stimuli are large enough, peripheral color vision is just like foveal vision.

Supplementary material

Discussion Size matters (Mullen et al. 2005)

“sinring” stimuli: Radial size was only 1.5 deg.

Introduction Cortical representation

Cortical magnification factor

The central part of the visual field (10 deg) is represented by about half of all neurons in primary visual cortex V1

Cone density in the retina (1D)

Introduction Physiology

• Size of the receptive fields

• Cortical representation

• Center – surround ratio

• Random wiring

• Results physiology – psychophysics

Discussion Biased sample? (Martin et al. 2001)

34/53 overt red-green response

28/35 cone-opponent

11 not significantly different from foveal cells

Results Cone-opponent thresholds

Further evidence for chromatic discrimination

Introduction Color opponency at larger eccentricities

Several factors can contribute to preserve color opponency

• selective wiring (vs. random wiring)

• elongated RFs

• unequal/random distribution of cone types at each retinal location

Introduction Physiology

„Random Wiring“ vs. „Selective Wiring“

Jusuf et al., 2006

Introduction Selectivity by elongated RFs

Martin et al., 2001

Midget RF centers are elongated and may rotate to sample one cone type more than the other to increase cone-purity

Introduction Elongated RFs can increase the L/(L+M) ratio

Martin et al., 2001

Methods Calibration