1. LAMINAR DEVELOPMENT How do V1 Ocular Dominance & Orientation maps develop? Ocular dominance columns and orientation tuning are found before eye opening in the cat How do cortical columns form? Consistent tuning in vertical V1 penetrations before interlaminar cortical connections have matured HYPOTHESIS: The Cortical Subplate Model of laminar development Subplate develops maps that are taught to LGN-to-layer 4 connections Subplate input guides layer 2/3 clustering Subplate input guides layers 4 and 2/3 connections Grinvald (http://www. weizmann.ac.il/brain/images/cubes.html) 2. ADULT ORGANIZATION OF V1 Ocular Dominance Columns (ODCs) Alternating stripes of cortex respond preferentially to visual inputs of each eye R/ L in Figure Orientation Columns A smooth pattern of changing orientation preference Organized in a pinwheel like fashion ( in Figure) Horizontal projections connect areas of same ocular dominance Horizontal projections connect areas of similar orientation preference 3. LATERAL CONNECTIONS IN V1 Clusters of lateral projections are found in the supragranular and infragranular layers Bosking et al., 1997 Retina Oriented Learns geniculocortical map Unoriented On Center / Off surround Learns corticogenicular map Random retinal activity drives network before eye opening Subplate LGN 7. MODEL SUBPLATE MAP Orientation Map Learned in Subplate Left Monocular Stripes Afferents compete for territory Monocular Layers Conserved # synapses Eyes are uncorrelated Subplate LGN Ocular Dominance Map Learned in Subplate Retina Right Subplate map is taught to Layer 4 Patterned visual inputs drive segregation of ON and OFF subfields in Layer 4 neurons At stimulus edges, ON and OFF thalamic cells are spatially out of phase and thus anti-correlated Left Right Retina LGN Subplate Layer 4 8. MODEL LAYER 4 MAP Oriented Learned Layer 4 map emulates LGN-Subplate map Subplate activity guides development of LGN to Layer 4 connections ON and OFF layers become equally active when eyes open When eyes open, random retinal activity is replaced by patterned visual inputs 9. MODEL LAYER 2/3 MAP Subplate guides horizontal connections Retina LGN Subplate Left Right Layer 2/3 Synapses with Subplate in Layer 1 Clustered horizontal connections Layer 4 develops connections to 2/3 Subplate activity instructs development of Layer 2/3 connections On Center / Off surround Random retinal activity drives network before eye opening Layer 4 10. MODEL LGN-TO-SUBPLATE MAP Orientation Columns Ocular Dominance Columns Ocular Dominance Columns develop in subplate with help of conservation of LGN synapses [plot of ocularity index] Oriented cells self organize in subplate [Length shows degree of orientation tuning and angle shows prefered orientation] 11. QUANTIFICATION OF SUBPLATE TUNING Matrices of Weights Feedforward pattern of connections from the LGN Receptive fields initialized as weight noise Oriented profiles develop Peak Orientation Network is probed with stimuli of different orientation Peak orientation is plotted Length shows orientation index Orientation Tuning Curves Orientation tuning curves calculated for each cell Most fit gaussian profile with halfwidths ~30˚ for tuned cells 0.79 1.00 0.54 0.48 12. LEARNED LGN-TO-4 MAP Subplate Layer 4 Ocular Dominance Columns Orientation Columns Maps in the Subplate are taught to Layer 4 13. RECIPROCAL CONNECTIONS BETWEEN LGN & SUBPLATE Oriented corticofugal connections match orientation of V1 cells LGN to Subplate Subplate to LGN Orientation same as LGN to Subplate connections Patterns are transferred to the later developing layer 6 to LGN connections Raw connection patterns Data from monocular OFF simulation Weights are clearly oriented Murphy et al., (1999) Model Results AFFERENTS FROM ON LAYERS AFFERENTS FROM OFF LAYERS 14. ON/OFF IN LAYER 4 ON and OFF segregation when eyes open Spatially offset ON & OFF cells fire to same edge Anti-correlation drives segregated receptive fields Significant diversity exists in arrangement of ON and OFF subfields in cortical receptive fields 15. MODEL SUBPLATE ABLATION Ocular Dominance Columns fail to develop Orientation Tuning fails to develop Some cells have a significant tuning index but inspection of the curves show no tuning (Ghosh & Shatz 1992) (Kanold et al., 2001) 16. LAYER 2/3 CLUSTERED HORIZONTAL CONNECTIONS Subplate Guides Horizontal Clusters Reciprocal connections between layer 2/3 cells Cluster precision due to local isotropic filters Size and spacing due to size of filters Noisy filters would result in more realistic pattern Model Results 17. LAYER 4-TO-2/3 CONNECTIONS Subplate Provides Vertical Correlations Focused connections contrast with 2/3 clusters Learning gated by layer 2/3 activity Connections develop without use of distance bias Model Results 18. BDNF AND SUBPLATE ABLATION Subplate ablation increases cortical BDNF Increase or Decrease of BDNF blocks ODC BDNF affects release of Glutamate and GABA Model insights: Parameters Matter Increasing excitation reduces the selectivity of cells by enlarging their receptive fields Increasing inhibition reduces activity and results in noisier receptive fields Equally changing both excitation and inhibition causes a shift in balance Increasing total input can reduce network activity since DOGs suppress uniform inputs (Ghosh & Shatz, 1994) (Cabelli et al., 1995, 1997) (Berardi & Maffei, 1999) 19 OTHER MODELS Most models are single layered Other developmental models have not demonstrated how vertical columns arise Clustering in Layer 2/3 rarely addressed The coordinated development of layer 2/3 horizontal connections with maps of orientation and ocular dominance has not previously been modeled The subplate goes unnoticed Our model is the first to use the subplate to explain cortical development LAMINART Model (Grossberg and Williamson 2001) Demonstrates how patterned vision can refine horizontal connections in Layer 2/3 After development these circuits generate properties of adult perceptual grouping The present model is consistent with these results 20. CONCLUSIONS Subplate enables development of Columns Learns Orientation and Ocular Dominance Maps Teaches OR and ODC to Layer 4 Instructs clustering of Layer 2/3 connections Guides growth of interlaminer connections Removal of Subplate eliminates map formation Eye opening segregates ON & OFF inputs The introduction of patterned vision provides a spatial anticorrelation of ON and OFF cells Future Directions Allow subplate to die Develop layer 2/3 maps Incorporate more realistic inhibitory circuit and develop inhibitory connections Generalizes model to other cortical areas, such as A1 and extrastriate areas 21. REFERENCES Albus , K., & Wolf, W. (1984). J Physiol , 348, 153-185. Allendoerfer , K. L., & Shatz , C. J. (1994). Annu Rev Neurosci , 17, 185-218. Bosking , W. H., Zhang, Y., Schofield, B., & Fitzpatrick, D. (1997). J Neurosci , 17(6), 2112-2127. Cook, P. M., Prusky , G., & Ramoa , A. S. (1999). Vis Neurosci , 16(3), 491-501. Crair, M. C., Horton, J. C., Antonini , A., & Stryker, M. P. (2001). J Comp Neurol , 430(2), 235-249. Ghosh , A., & Shatz , C. J. (1992).. Science, 255(5050), 1441-1443. Ghosh , A., & Shatz , C. J. (1994). J Neurosci , 14(6), 3862-3880 Grossberg, S., & Williamson, J. R. (2001). Cereb Cortex, 11(1), 37-58. Kanold , P.O., Kara, P., Reid, R.C., Shatz , C.J. (2001). Soc. Neurosci . Abstr ., P27.16. Murphy P.C., Duckett S.G., Sillito A.M.(1999). Science, 286, 1152-1154. Yoshioka, T., Blasdel , G. G., Levitt , J. B., & Lund, J. S. (1996). Cereb Cortex, 6(2), 297-310. S upported in part by AFOSR, DARPA, NSF, and ONR. Crair et al., 2001 4. VISUAL DEVELOPMENT TIMELINE OFF field bias Cortical cells respond to negative contrast stimuli 76% at P8 Contralateral Bias At P8 contralateral eye dominates Horizontal Connections ODCs project more to areas of same dominance than other dominance Iso-oriented areas project primarily to areas of similar orientation ~20-30% of projections to different “modules” Visual Cortex Lateral Geniculate Nucleus E37 E44 E51 E58 E65/P0 P7 P14 P21 P28 LGN axons in Subplate LGN axons in cortical plate (layers 5 and 6) LGN axons in Layer 4 Ocular dominance columns emerge Orientation selective neurons detected Critical period for visual deprivation Retinal ganglion cell afferents in LGN Retinogeniculate afferents segregate LGN cytoarchitechtonic lamina differentiate Albus & Wolf 1984 Crair et al., 2001 Molecular gradients provide a crude map Epherins, Netrins, Semaphorins, Slit, etc. Activity refines existing maps Activity provides correlations between related cells Retina Tectum Nasal Temporal Anterior Posterior A Nasal-Temporal gradient of Eph receptors in retinal ganglion cells interact with anterior to posterior gradients of repellant epherin ligand in the Tectum. The result is a crude retinotopic map. Normal Development ‡ Binocular Deprivation ‡ Monocular Deprivation ‡ Cook et al., 1999 LGN 5. DEVELOPMENTAL MECHANISMS TTX 6. CORTICAL SUBPLATE Cortical subplate forms circuit with LGN LGN afferents wait in subplate for weeks before they synapse in cortical plate (Ghosh & Shatz 1994) Ablation of subplate stops map formation When subplate is ablated, ocular dominance columns fail to form (Ghosh & Shatz 1992) Orientation tuning and orientation maps do not develop after subplate ablation (Kanold et al., 2001) Subplate connects to Layer 4 (Ghosh 1995) Connections exist from subplate to Layer 4 when LGN afferents begin to grow into Layer 4 Subplate connects to Layer 2/3 Subplate connects to Layer 1 (Allendoerfer & Shatz 1994) Layer 2/3 has dendritic arborizations in Layer 1 Migrating Neuron H o w D o L a m i n a r C i r c u i t s D e v e l o p ? T h e R o l e o f t h e C o r t i c a l S u b p l a t e i n t h e D e v e l o p m e n t a n d L a m i n a r C o o r d i n a t i o n o f O r i e n t a t i o n a n d O c u l a r D o m i n a n c e M a p s i n V 1 A a r o n S e i t z a n d S t e p h e n G r o s s b e r g D e p a r t m e n t o f C o g n i t i v e a n d N e u r a l S y s t e m s , B o s t o n U n i v e r s i t y , B o s t o n , M A a s e i t z @ b u . e d u s t e v e @ b u . e d u h t t p : / / w w w . c n s . b u . e d u / ~ a s e i t z