K. Miller, J. Keller, and M. Stryker. Ocular dominance column development: analysis and simulation. Science, 245:605615, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....recent modelling efforts have concentrated on the formation of ocular dominance and orientation columns, but some have included map formation as well. Obermayer [51] uses a variant of the Kohonen algorithm which demonstrates map formation using a biological input representation. Miller s model [47] displays activitydependent map refinement, but the initial topography is fully specified, and axonal arbors narrowover time around a fixed location. 4.3 The Model The model described in this paper is based on the following (biologically motivated) description of the developmental process: ....

Ken Miller. Ocular dominance column development: Analysis and simulation. Science, 245, 1989.

....are of a mexican hat type: positive on short distances, negative on somewhat longer distances and vanishing for large distances. This resulted in a clustering of neurons that are sensitive to bars of almost the same orientation. A feature that is also found in cats. Miller, Keller and Stryker [28] used a model very similar to Linsker s to explain the onset of occular dominance patches in many adult mammals. These patches are groups of cells that, although initially all are receiving input from both eyes, have specialized in processing information from just on eye. These and other results ....

K.D. Miller, J.B. Keller, and M.P. Stryker. Ocular dominance column development: analysis and simulation. Science, 245:605--615, 1989.

....have yet to be fully captured in a model. 9.3.1 Hebbian Learning The Hebbian model of topography described in section 9.2. 2 incorporates aspects (such as subtractive normalisation) that were shown in chapter 8 to be sufficient to create ocular dominance when there is input from two eyes (Miller, Keller Stryker, 1989). Figure 9.8A shows the initial weights from input to output, for which the two eyes are statistically identical. We refer to inputs as being from the eyes, although they are really from relay cells in the eye specific layers of the LGN. Figure 9.8D shows covariance functions between all the input ....

Miller, KD, Keller, JB & Stryker, MP (1989) Ocular dominance column development: Analysis and simulation. Science 245:605-615.

....byBox, Head Neck and Cylinder were low) the activation vectors did overlap somewhat for these three inputs (for example, unit 10 in Table 5.1) This Feature Detector, whichwas trained on random inputs (see Section 3. 2) was intended to emulate the development of the early visual system in cats (Miller et al. 1989) and monkeys (Mooney et al. 1996# Shatz, 1994) via spontaneous activation. In the end, however, it was used uniformly among all the networks, which seems unrealistic: each individual ought to develop its own feature detector. For this first stab at modeling the predisposition seriously,however, ....

.... only some of the weights in the artificial genome (with the rest random) and trying again to construct a predisposition that includes some sort of self organization, analogous to (but later than) the self organization of early visual feature detectors in primates and cats (Mooney et al. 1996# Miller et al. 1989). 5.2.2 The Predisposition as an Attentional Mechanism Horn and his colleagues (Horn, 1985# Bateson and Horn, 1994) havecharacterized the predisposition as an attentional mechanism. The model in Bateson and Horn (1994) suggests that this attentional mechanism might be extremely low level, since ....

K. Miller, J. Keller, and M. Stryker. Ocular dominance column development: analysis and simulation. Science, 245:605--615, 1989.

.... characterizes the properties of long term potentiation and depression (LTP LTD) in the cortex (e.g. Bear, 1996) Psychologically, inhibition is important for modeling attentional and other phenomena, and Hebbian learning has been used in modeling a number of different learning phenomena (e.g. Miller, Keller, Stryker, 1989). In the context of generalization in interactive networks, the specific biases of inhibitory competition and Hebbian learning constrain both the interactive activation dynamics (in the case of inhibitory competition) and the learned weight patterns such that the network actually produces ....

Miller, K. D., Keller, J. B., & Stryker, M. P. (1989). Ocular dominance column development: Analysis and simulation. Science, 245, 605--615.

.... ways of implementing the Hebbian or the competitive parts of the rule, different models for the activity itself (either random or somehow patterned [41] from which the development arises, and different degrees of abstraction from the neural basis, all lead to apparently different algorithms (eg [42, 43, 44, 45, 46]) 5 and there is as yet not enough experimental data to distinguish between those that compete. Also, they have almost exclusively been confined to the primary sensory areas; cortical mechanisms are not apparently different deeper in, but it is not clear that these algorithms can be sensibly ....

Miller, KD, Keller, JB and Stryker, MP. Ocular dominance column development: Analysis and simulation. Science, 1989 245:605-15.

.... have been demonstrated and a number of studies addressed these phenomena [Goodhill, 1997] Many of the different learning rules proposed are actually derived from similar principles [Erwin et al. 1995] For example, the competition for a substrate together with the use of local correlation [Miller et al. 1989] leads to patterns as those observed in the mammalian visual cortex. Nevertheless, a number of problems remain, e.g. the relation of patterns determined by different features such as orientation vs. ocular dominance [Erwin et al. 1995] and the global arrangement of these structures within a ....

Miller, K. D., Keller, J. B., & Stryker, M. P. (1989). Ocular dominance column development: analysis and simulation. Science, 245, 605-615.

.... tuning remain largely intact [4, 31] Both orientation and direction selectivity are sharpened by normal visual experience; for reviews see [22, 47, 52, 58] Numerous models have been formulated to describe activity dependent learning as a self organization process in accord with a Hebbian rule [16,17,35 37,47,51,63,65,67]. Most of these models have exclusively concentrated on the development of spatial properties of the receptive field, such as orientation selectivity. In particular, in the model of one of the present authors [16, 47] orientation selectivity develops through a competition between ON center and ....

Miller KD, Keller JB, Stryker MP (1989) Ocular dominance column development: Analysis and simulation. Science 245:605--615

....lateral intralayer interaction, the feedforward synaptic arbor density function, and parallel ON , OFF arrays are all thought to contribute to the 36 symmetry breaking mechanism by which oriented aRFs and orientation dominance maps can arise. A considerable number of analyses and simulations [3, 4, 8, 9, 19, 28, 29, 30, 31, 32, 33, 38, 41] on the self organization in visual system have recently expanded Linsker s ideas by introducing more biologically motivated factors, and in particular, by including lateral intralayer interactions. Our approach can be adapted to explicitly analyze the role played by the lateral intralayer ....

Miller, K.D., Keller, J.B., & Stryker, M.P. (1989). Ocular dominance column development: Analysis and simulation. Science, 245, 605-615.

....horizontal connections in the primary visual cortex. 1 Introduction Many areas of the neocortex form two dimensional topographic maps of peripheral senses. Such maps are believed to develop in a self organizing process based on cooperation and competition between neurons (Hubel and Wiesel 1965; Miller et al. 1989; von der Malsburg 1973; von der Malsburg and Singer 1988) The Self Organizing Feature Map (SOFM) algorithm (Kohonen 1989, 1990) is an abstract computational model of this process. At each step during self organization, the algorithm finds the neuron whose input synaptic weights are closest to ....

Miller, K. D., Keller, J. B., and Stryker, M. P. (1989). Ocular dominance column development: Analysis and simulation. Science, 245:605--615.

....higher dimensions, where the added dimensions represent other attributes such as which eye the input comes from etc. The inputs in all of these models intend to model normal rearing, i.e. rearing in an environment composed of objects and patterns. Another family of models (Linsker, 1986; Miller et al. 1989; Miller, 1994) has attempted to explain the receptive field structures which form before eye opening. In these models it is assumed that plasticity is driven by noise originating in retinal neurons, even in the absence of a patterned environment. Another feature common to these models is that ....

....the two dimensional retinal sheet before eye opening. Most of the aforementioned models have attempted to show how the organization of receptive fields could emerge from plasticity (von der Malsburg, 1973; Nass and Cooper, 1975; Kohonen, 1982; Scofield and Cooper, 1985; Obermayer et al. 1992; Miller et al. 1989; 8 Inputs Learning Rule Interactions Model Symbolic Hebb Network Kohonen 82 (Low dimensional) jL 2 j Non Linear Obermayer Et. Al 92 Normalization Simplified Hebb jL 1 j Network, Linear Malsburg 73 Hebb Single Cell Nass and Cooper 75 fixed threshold Network BCM Single Cell BCM 82 ....

[Article contains additional citation context not shown here]

Miller, K. D., Keller, J. B., and Striker, M. P. (1989). Ocular dominance column development: Analysis and simulation. Science, 245:605--615.

.... can be resolved by assuming that while Hebb type plasticity is responsible for developmental synaptic changes, the organization of presynaptic activity which exists under conditions of visual deprivation is sufficient to drive the neurons towards greater specificity (Linsker 1986a c; Miller 1989, 1992; Miller et al. 1989) As a reverse Hebb rule enhances response specificity by balancing the push pull system of ONand OFF center afferents, the sufficient condition is that the activity of ON and OFF center retinal ganglion cells be negatively correlated, a condition which will be met ....

.... by assuming that while Hebb type plasticity is responsible for developmental synaptic changes, the organization of presynaptic activity which exists under conditions of visual deprivation is sufficient to drive the neurons towards greater specificity (Linsker 1986a c; Miller 1989, 1992; Miller et al. 1989). As a reverse Hebb rule enhances response specificity by balancing the push pull system of ONand OFF center afferents, the sufficient condition is that the activity of ON and OFF center retinal ganglion cells be negatively correlated, a condition which will be met by diffuse illumination as ....

[Article contains additional citation context not shown here]

Miller, K. D., Keller, J. B. & Stryker, M. P. (1989) Ocular dominance column development: Analysis and simulation. Science 245, 605-615.

....the one proposed by Oja (1982) This rule extracts the first principal component of the input environment. However this analysis is relevant to a whole family of Hebbian type learning rules, since the receptive fields of several other variants (von der Malsburg, 1973; Linsker, 1986; Linsker, 88; Miller et al. 1989; Miller, 1994, eg: have been shown (Linsker, 88; Miller and MacKay, 1994) to governed by the first Principal component. In this paper we model this preprocessing by a convolution of the natural scenes with a center surround filter. We force the receptive fields to be localized by assuming that ....

Miller, K. D., Keller, J. B., and Striker, M. P. (1989). Ocular dominance column development: Analysis and simulation. Science, 245:605--615.

....for a recent review) Both classes of hypotheses are based on activity driven Hebbian learning for the connection strengths of the afferent fibers under certain constraints. According to the first hypothesis, the so called correlation based learning hypothesis [Linsker, 1986, Yuille et al. 1989, Miller et al. 1989, Tanaka, 1990, Miller, 1994] neural response properties are determined by the first and second moments of the distribution of afferent activity patterns, which are present at the time when the cortical response properties develop. According to the second hypothesis, the so called competitive ....

....j (S jy ) 2 = const. exactly or by soft approximations to the hard constraint that take the form of biologically more appealing synaptic decay terms. The influence of subtractive constraints and weight clipping is discussed in section 6. For the biological interpretation we refer the reader to [Miller et al. 1989, Miller, 1994, Obermayer et al. 1990c] however, two comments need to be made here. Firstly, the soft max normalization of the neurons outputs in eq. 1) implicitely assumes non local lateral interactions. For the purpose of this study one may think of eq. 1) as to apply to a small region in ....

[Article contains additional citation context not shown here]

Miller, K., Keller, J. B., and Stryker, M. P. (1989). Ocular dominance column development: Analysis and simulation. Science, 245:605--615.

....neurons. The learning rule assumed herein is a normalized Hebbian rule(Oja, 1982; Miller and MacKay, 1994) The type of network architecture assumed here is similar to that used in many other synaptic models of cortical plasticity (von der Malsburg, 1973; Nass and Cooper, 1975; Linsker, 1986b; Miller et al. 1989; Miller, 1994, etc. Rather then presenting a new model we have attempted, in this paper, to analyze the behavior of these existing models, and map them onto much simpler models that we might be able to understand better. When this analysis is be performed, it becomes clear that in order to ....

Miller, K. D., Keller, J. B., and Striker, M. P. (1989). Ocular dominance column development: Analysis and simulation. Science, 245:605--615.

No context found.

K. Miller, J. Keller, and M. Stryker. Ocular dominance column development: analysis and simulation. Science, 245:605615, 1989.

No context found.

Miller, K. D., Keller, J. B., and Stryker, M. P. (1989) Ocular dominance column development: Analysis and simulation. Science 245, 605--615.

No context found.

K.D. Miller, J.B. Keller, and M.P. Stryker. Ocular dominance column development: Analysis and simulation. Science, 245:605 615, 1989.

No context found.

K. D. Miller, J. B. Keller, and M. P. Stryker. Ocular dominance column development: Analysis and simulation. Science, 245:605--615, August 1989.

No context found.

Miller, K.D., Keller, J.B. and Stryker, M.P. (1989) Ocular dominance column development: analysis and simulation Science 245, 605--615

No context found.

Miller, K.D., Keller, J.B. and Stryker, M.P. (1989) Ocular dominance column development: analysis and simulation Science 245, 605--615

No context found.

Miller, K. D., Keller, J. B., and Stryker M. P. 1989. Ocular dominance column development: Analysis and simulation. Science, 245:605-615.

No context found.

Miller, K. D., Keller, J., and Stryker, M. P. (1989). Ocular dominance column development: Analysis and simulation. Science, 240:605--615.

No context found.

Miller, K.D., Keller, J.B., & Stryker, M.P. (1989) Ocular dominance column development: Analysis and simulation. Science, 245, 605--615.

No context found.

Miller, K.D., Keller, J.B., & Stryker, M.P. (1989) Ocular dominance column development: Analysis and simulation. Science, 245, 605--615.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC