RIKEN BSI > RIKEN BSI News > No. 29 (Aug. 2005) > Research Results at BSI
RIKEN BSI News No. 29 (Aug. 2005)

Language: English » Japanese

Research Results at BSI

Development of Orientation Maps: Nature or Nurture

Laboratory for Visual Neurocomputing


Neurons in the visual cortices of monkeys and cats selectively respond to stimuli, which arrange themselves into orientation maps. For over thirty years researchers have sought to learn precisely how the maps emerge during development. But the answers are elusive. We still do not know whether intrinsic mechanisms or visual experience drive map formation.


Fig.1: Orientation maps obtained from the self-organization model under different stimulus environments
a) Orientation angle maps (preferred orientations indicated by colors), b) orientation polar maps (preferred orientation and orientation selectivity are indicated by colors and brightness, respectively), c) histograms of preferred orientations, and d) power spectra of the arrangement of orientation preferences, for different stimulus conditions in which moving gratings of various orientations are presented to the model retina (1); spontaneous activity is generated in the model LGN with no visual input to the retina (2); a single, oriented moving grating is presented to the retina (3); a single, oriented moving grating is presented to the retina while spontaneous activity is generated in the LGN (4).


Fig.2: Orientation maps reconstructed by intrinsic signal optical imaging of area 18 of kittens reared under different visual conditions
a) Orientation polar maps, b) histograms of preferred orientations, and c) power spectra of the arrangement of orientation preferences, for a kitten reared in a normal visual environment (1), a kitten reared in darkness (2), a kitten reared with cylindrical-lens-fitted goggles for vertical orientation exposure (3), a kitten reared with similar goggles for horizontal orientation exposure (4), a kitten reared with the goggles for vertical orientation exposure for 2 hours a day, otherwise kept in darkness (5).

To determine the extent of visual experience's influence on orientation map formation, we conducted computer simulations using our activity-dependent self-organization model. We assumed that: (1) moving gratings of various orientations are presented to the model retina; (2) spontaneous activity occurs the model lateral geniculate nucleus (LGN) in the absence of visual exposure; (3) a single, oriented moving grating is presented to the model retina; and (4) a single, oriented grating is presented with spontaneous activity in the model LGN. Normal and dark rearing as well as rearing with single orientation exposure alone or mixed with dark rearing episodes were considered. Fig. 1 shows the orientation angle map, orientation polar map, orientation histogram and power spectrum obtained from computer simulation for each condition. The simulation results suggest that orientation maps form even in dark rearing conditions, where selectivity is low. Over-representation of the exposed orientation resulted from single orientation exposure, while that over-representation was suppressed when exposure was interlaced with dark rearing episodes. To demonstrate theoretical predictions, we then conducted intrinsic signal optical imaging experiments using kittens reared: (1) under normal visual conditions; (2) in complete darkness; (3) with continuous exposure to vertical orientation through chronically mountable cylindrical-lens-fitted goggles; (4) with continuous exposure to horizontal orientation through goggles; and (5) with vertical orientation exposure through goggles for 2 hours a day or otherwise in darkness. The reconstructed orientation polar map, orientation histogram and power spectrum for each condition are shown in Fig. 2. Experimental results confirm theoretical predictions.


In previous experiments, animals experienced single orientations either in a striped environment or through mask-type goggles that could be accidentally removed during manipulation. This would skew data. This time the goggles were stably fixed on the animal's skull using a head brace and extreme over-representations were observed. Therefore, orientation maps can be drastically reorganized when visual orientation is restricted during the critical period and spontaneous activity in the LGN can suppress the over-representation of the exposed orientation.


Shigeru Tanaka, Masanobu Miyashita and Jérôme Ribot: Neural Networks Volume 17, 1037-1389 (2004)

Published by

  • RIKEN Brain Science Institute
    Brain Science Promotion Division
    2-1 Hirosawa, Wako, Saitama, 351-0198 JAPAN
    Tel: +81 48 462 1111
    Facsimile: +81 48 462 4914
    Email: bsi@riken.jp
  • All copyrights reserved and protected by Japanese and International Copyright Law.