Jean-Luc Stevens PhD

Jean-Luc Stevens


Research Interests

My primary research interest is in how computational models may be used to understand the principles underlying cortical intelligence and how the neocortex processes natural language and performs deliberative planning. The first step towards this goal is to build a a general, cortical model that can be applied to visual processing to try and account for the predictive and contextual phenomena observed in the primary visual cortex. In particular, a developmental model is needed to explain how the neocortex encodes spatiotemporal statistical expectations via Hebbian learning and how the cortical connectivity enables contextual modulation and predictive processing of spatiotemporal sequences.

Publications:
2014
  A model relating temporal processing across spatial and temporal scales using electrophysiological and optical imaging data in primate V1
Stevens, J-L, Chemla, S, Benvenuti, G, Chavane, F & Bednar, J 2014, 'A model relating temporal processing across spatial and temporal scales using electrophysiological and optical imaging data in primate V1' 44th Society for Neuroscience Annual Meeting, Washington, DC, United States, 15/11/14 - 19/11/14, .
General Information
Organisations: Institute for Adaptive and Neural Computation .
Authors: Stevens, Jean-Luc, Chemla, Sandrine, Benvenuti, Giacomo, Chavane, Frederic & Bednar, James.
Publication Date: 2014
Publication Information
Category: Poster
Original Language: English
  Relationship between natural image statistics and lateral connectivity in the primary visual cortex
Rudiger, P, Stevens, J-L, Talluri, B & Bednar, JA 2014, 'Relationship between natural image statistics and lateral connectivity in the primary visual cortex' Computational and Systems Neuroscience (Cosyne) 2014, Salt Lake City, United States, 27/02/14 - 2/03/14, pp. 148-149.
The distribution of orientation-selective cells in the primary visual cortex has been found to reflect the first-order statistics of visual inputs, i.e. which orientations are most common during a critical period [1]. Similarly, some properties of the lateral connections between these cells have been found to reflect the second-order statistics of images, i.e. the coocurrence statistics of oriented edge elements, but the results have differed by species. Specifically, horizontal connections have been found to be elongated along the axis of preferred orientation in tree shrew [2] and owl monkey [3], but not in macaque [4]. It is unclear whether these results indicate genuine species differences, or perhaps differences in the visual environments in which these animals were raised. Here we analyse the effect of input statistics on lateral excitatory connectivity in a developmental model of primary visual cortex, by relating differences in co-occurence statistics of distinct image datasets, analysed using methods presented in [5], to corresponding differences in the emergent long-range lateral structure. The goal is to propose a testable series of mechanisms linking visual inputs, to cortical structure, and then to predicted behavioral results on tests of surround modulation. The model, based on earlier variations of a self-organizing map model, develops robust yet adaptive orientation maps [6], long-range lateral connectivity, and a wide range of contextual modulatory effects in a spatially calibrated circuit of macaque V1 [7]. By training the model using images with previously characterized second-order statistics, we demonstrate that the co-occurence of oriented edges is encoded within the horizontal connectivity of our model. On that basis we predict patterns of lateral connectivity for animas reared in different environments, which can be tested by in future experiments.
[1] Shigeru Tanaka, Jerome Ribot, Kazuyuki Imamura, and Toshiki Tani. Orientation-restricted continuous visual exposure induces marked reorganization of orientation maps in early life. NeuroImage, 30(2):462?77, April 2006.
[2] W H Bosking, Y Zhang, B Schofield, and D Fitzpatrick. Orientation selectivity and the arrangement of horizontal connections in tree shrew striate cortex. The Journal of Neuroscience, 17(6):2112?27, March 1997.
[3] L C Sincich and G G Blasdel. Oriented axon projections in primary visual cortex of the monkey. The Journal of Neuroscience, 21(12):4416?26, June 2001.
[4] Alessandra Angelucci, Jonathan B Levitt, Emma J S Walton, Jean-Michel Hupe, Jean Bullier, and Jennifer S Lund. Circuits for local and global signal integration in primary visual cortex. The Journal of Neuroscience, 22(19):8633?46, October 2002.
[5] Laurent Perrinet and James A. Bednar. Edge co-occurrences are sufficient to categorize natural versus animal images. Submitted.
[6] Jean-luc R Stevens, Judith S Law, Jan Antolik, and James A Bednar. Mechanisms for stable, robust,and adaptive development of orientation maps in the primary visual cortex. Journal of Neuroscience,33(40):15747?15766, 2013.
[7] Philipp Rudiger and James A. Bednar. Unifying anatomical, psychophysical and developmental circuit models of primary visual cortex. In Society for Neuroscience, page Program No. 639.11/EE10, 2013.
General Information
Organisations: Institute for Adaptive and Neural Computation .
Authors: Rudiger, Philipp, Stevens, Jean-Luc, Talluri, Bharath & Bednar, James A..
Number of pages: 1
Publication Date: 2014
Publication Information
Category: Poster
Original Language: English
2013
  Mechanisms for Stable, Robust, and Adaptive Development of Orientation Maps in the Primary Visual Cortex
Stevens, J-L, Law, J, Antolik, J & Bednar, JA 2013, 'Mechanisms for Stable, Robust, and Adaptive Development of Orientation Maps in the Primary Visual Cortex' Journal of Neuroscience, vol 33, no. 40, pp. 15747-15766. DOI: 10.1523/JNEUROSCI.1037-13.2013
Development of orientation maps in ferret and cat primary visual cortex (V1) has been shown to be stable, in that the earliest measurable maps are similar in form to the eventual adult map, robust, in that similar maps develop in both dark rearing and in a variety of normal visual environments, and yet adaptive, in that the final map pattern reflects the statistics of the specific visual environment. How can these three properties be reconciled? Using mechanistic models of the development of neural connectivity in V1, we show for the first time that realistic stable, robust, and adaptive map development can be achieved by including two low-level mechanisms originally motivated from single-neuron results. Specifically, contrast-gain control in the retinal ganglion cells and the lateral geniculate nucleus reduces variation in the presynaptic drive due to differences in input patterns, while homeostatic plasticity of V1 neuron excitability reduces the postsynaptic variability in firing rates. Together these two mechanisms, thought to be applicable across sensory systems in general, lead to biological maps that develop stably and robustly, yet adapt to the visual environment. The modeling results suggest that topographic map stability is a natural outcome of low-level processes of adaptation and normalization. The resulting model is more realistic, simpler, and far more robust, and is thus a good starting point for future studies of cortical map development.
General Information
Organisations: Institute for Adaptive and Neural Computation .
Authors: Stevens, Jean-Luc, Law, Judith, Antolik, Jan & Bednar, James A. .
Number of pages: 20
Pages: 15747-15766
Publication Date: 2 Oct 2013
Publication Information
Category: Article
Journal: Journal of Neuroscience
Volume: 33
Issue number: 40
ISSN: 0270-6474
Original Language: English
DOIs: 10.1523/JNEUROSCI.1037-13.2013
  An automated and reproducible workflow for running and analyzing neural simulations using Lancet and IPython Notebook
Stevens, J-L, Elver, M & Bednar, JA 2013, 'An automated and reproducible workflow for running and analyzing neural simulations using Lancet and IPython Notebook' Frontiers in Neuroinformatics, vol 7.
Lancet is a new, simulator-independent Python utility for succinctly specifying, launching, and collating results from large batches of interrelated computationally demanding program runs. This paper demonstrates how to combine Lancet with IPython Notebook to provide a flexible, lightweight, and agile workflow for fully reproducible scientific research. This informal and pragmatic approach uses IPython Notebook to capture the steps in a scientific computation as it is gradually automated and made ready for publication, without mandating the use of any separate application that can constrain scientific exploration and innovation. The resulting notebook concisely records each step involved in even very complex computational processes that led to a particular figure or numerical result, allowing the complete chain of events to be replicated automatically. Lancet was originally designed to help solve problems in computational neuroscience, such as analyzing the sensitivity of a complex simulation to various parameters, or collecting the results from multiple runs with different random starting points. However, because it is never possible to know in advance what tools might be required in future tasks, Lancet has been designed to be completely general, supporting any type of program as long as it can be launched as a process and can return output in the form of files. For instance, Lancet is also heavily used by one of the authors in a separate research group for launching batches of microprocessor simulations. This general design will allow Lancet to continue supporting a given research project even as the underlying approaches and tools change.
General Information
Organisations: Institute for Adaptive and Neural Computation .
Authors: Stevens, Jean-Luc, Elver, Marco & Bednar, James A..
Publication Date: 2013
Publication Information
Category: Article
Journal: Frontiers in Neuroinformatics
Volume: 7
ISSN: 1662-5196
Original Language: English
  An automated simulator-independent workflow for reproducible simulation and analysis using Lancet and IPython Notebook
Stevens, J-L, Elver, M & Bednar, J 2013, 'An automated simulator-independent workflow for reproducible simulation and analysis using Lancet and IPython Notebook' 22nd Annual Computational Neuroscience Meeting: CNS 2013, Paris, France, 13/07/13 - 18/07/13, . DOI: 10.1186/1471-2202-14-S1-P22
General Information
Organisations: Institute for Adaptive and Neural Computation .
Authors: Stevens, Jean-Luc, Elver, Marco & Bednar, James.
Number of pages: 2
Publication Date: 2013
Publication Information
Category: Poster
Original Language: English
DOIs: 10.1186/1471-2202-14-S1-P22
2012
  A temporal model of neural activity and VSD response in V1
Stevens, J-L & Bednar, J 2012, 'A temporal model of neural activity and VSD response in V1' 21st Annual Computational Neurscience Meeting: CNS 2012, Decatur, United States, 21/07/12 - 26/07/12, .
Mechanistic developmental models of the primary visual cortex (V1) in mammals have been able to replicate many of the large-scale spatial features of V1 neurons from experiments, such as their spatial receptive fields and the spatial organization into maps in V1 (reviewed in ref [1]). However, the models have previously been formulated at a very abstract level that does not account for the detailed, transient time course of neural responses. Conversely, there are a number of detailed, large-scale spiking models of the adult visual cortex, but these have not explained the development of feature preferences and feature maps, relying instead on prespecified patterns of connectivity. Here we present a new temporally and spatially calibrated model of cortical activity using ratebased units that could help unify these different types of explanation and levels of modelling. The model is called TCAL (Temporally CALibrated), and is a small variant on the GCAL model from the LISSOM family [1].
Compared to GCAL, the only change to the model mechanisms is to add hysteresis to the model LGN and V1 units. Hysteresis allows the damping of temporal responses to be controlled with one time-constant parameter per sheet. These two new parameters were set first for the LGN and then for V1to match results from electrophysiological recordings. Both onset and offset responses are matched against experimentally recorded peristimulus time histograms (PSTHs) for LGN [2] and cortical [3] neurons using the Invariant Response Description model. Despite the two orders of magnitude difference in time scales between GCAL and TCAL and the minimal change to the GCAL rate-based mechanisms, the time course of responses is already a remarkably close match (see Figure 1).
The transient responses of LGN neurons in TCAL are due to lateral inhibition between LGN/RGC cells already present in GCAL, but originally for purposes of contrast gain control. The transient responses at the V1 level are partly inherited from LGN but also reflect lateral connectivity in V1, originally for the purposes of map development. TCAL thus shows how the observed transient response properties can arise from the same mechanisms that lead to map development in mechanistic Hebbian models [1]. After calibrating the afferent response delays and distance-dependent lateral connection delays, TCAL can now be used to predict the spatial and temporal time course of voltage-sensitive-dye (VSD) responses to spatiotemporal visual stimuli, and allows the detailed dynamics of perceptual phenomena to be studied even without spiking models.

References
1. Miikkulainen R, Bednar JA, Choe Y, Sirosh J: Computational Maps in the Visual Cortex. Springer; 2005.
2. Funke K, Wörgötter F: On the significance of temporally structured
activity in the dorsal lateral geniculate nucleus (LGN). Progress in
Neurobiology 1997, 53:67-119.
3. Albrecht DG, Geisler WS, Frazor RA, Crane AM: Visual cortex neurons of monkeys and cats: Temporal dynamics of the contrast response
function. J. Neurophysiology 2002, 88:888-913.
General Information
Organisations: Institute for Adaptive and Neural Computation .
Authors: Stevens, Jean-Luc & Bednar, James.
Publication Date: 2012
Publication Information
Category: Poster
Original Language: English
  A spatio-temporally calibrated model of feature map development and neuronal activity in primate V1
Rudiger, P, Stevens, J-L & Bednar, J 2012, 'A spatio-temporally calibrated model of feature map development and neuronal activity in primate V1' Society for Neuroscience Annual Meeting 2012, New Orleans, 13/10/12 - 17/10/12, .
Detailed spiking models of the primary visual cortex (V1) in mammals use experimentally derived patterns of connectivity from adult animals, failing to explain how this connectivity develops or how it relates to function. In contrast, developmental models explore how connectivity arises as receptive fields and feature maps organize [1]. A unifying model would retain the parsimony of the developmental approach, while still accounting for the spatial and temporal characteristics of experimentally recorded activity.
Here we present the Spatio-Temporally CALibrated (S-TCAL) rate-based model of feature map development and cortical activity at parafoveal eccentricity of 4-5° in macaque. This model is a variant on GCAL [2] with spatially calibrated connectivity and temporal activity profiles in the lateral geniculate nucleus (LGN) and V1. The aim is to determine whether dynamics of V1 activity can be explained with a rate based model, using connectivity organised through visual experience without simulating spiking potentials.

[1] Geofrrey J. Goodhill. Contributions of Theoretical Modeling to the Understanding of Neural Map Development. Neuron Review 56:301-310, 2007.
[2] James A. Bednar. Building a mechanistic model of the development and function of the primary visual cortex. Journal of Physiology (Paris), 2012. In press.
General Information
Organisations: Institute for Adaptive and Neural Computation .
Authors: Rudiger, Philipp, Stevens, Jean-Luc & Bednar, James.
Publication Date: 2012
Publication Information
Category: Poster
Original Language: English

Projects:
What role does self-organized lateral connectivity in V1 play in predicting spatiotemporal response? (PhD)