Limits...
Towards reproducible descriptions of neuronal network models.

Nordlie E, Gewaltig MO, Plesser HE - PLoS Comput. Biol. (2009)

Bottom Line: Progress in science depends on the effective exchange of ideas among scientists.This hinders the critical evaluation of network models as well as their re-use.We believe that the good model description practice proposed here, together with a number of other recent initiatives on data-, model-, and software-sharing, may lead to a deeper and more fruitful exchange of ideas among computational neuroscientists in years to come.

View Article: PubMed Central - PubMed

Affiliation: Department of Mathematical Sciences and Technology, Norwegian University of Life Sciences, Aas, Norway.

ABSTRACT
Progress in science depends on the effective exchange of ideas among scientists. New ideas can be assessed and criticized in a meaningful manner only if they are formulated precisely. This applies to simulation studies as well as to experiments and theories. But after more than 50 years of neuronal network simulations, we still lack a clear and common understanding of the role of computational models in neuroscience as well as established practices for describing network models in publications. This hinders the critical evaluation of network models as well as their re-use. We analyze here 14 research papers proposing neuronal network models of different complexity and find widely varying approaches to model descriptions, with regard to both the means of description and the ordering and placement of material. We further observe great variation in the graphical representation of networks and the notation used in equations. Based on our observations, we propose a good model description practice, composed of guidelines for the organization of publications, a checklist for model descriptions, templates for tables presenting model structure, and guidelines for diagrams of networks. The main purpose of this good practice is to trigger a debate about the communication of neuronal network models in a manner comprehensible to humans, as opposed to machine-readable model description languages. We believe that the good model description practice proposed here, together with a number of other recent initiatives on data-, model-, and software-sharing, may lead to a deeper and more fruitful exchange of ideas among computational neuroscientists in years to come. We further hope that work on standardized ways of describing--and thinking about--complex neuronal networks will lead the scientific community to a clearer understanding of high-level concepts in network dynamics, and will thus lead to deeper insights into the function of the brain.

Show MeSH

Related in: MedlinePlus

Interpretation of Lumer [10] model architecture.The most parsimonious interpretation of the description of the primary visual cortical area Vp given by Lumer et al, is as two layers of 40×40 topographic elements, representing horizontal and vertical orientations, respectively.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2713426&req=5

pcbi-1000456-g003: Interpretation of Lumer [10] model architecture.The most parsimonious interpretation of the description of the primary visual cortical area Vp given by Lumer et al, is as two layers of 40×40 topographic elements, representing horizontal and vertical orientations, respectively.

Mentions: The most detailed explicit model studied here is the thalamocortical model presented by Lumer et al. [10]. The description of the cortical areas in this model (Vp and Vs), while complete, lacks in our opinion the clarity desirable of a good model description, and may thus help to identify rules for ideal model descriptions. For one, discussions on model design and properties are embedded in the model description, e.g., the reduction of a total of 32 “combinations of response selectivities” to just two included in the model, and a comparison of the number of neurons in the model to that found in animals. We believe that design decisions and model review should be kept separate from the model description proper for the sake of clarity, since they are independent intellectual endeavours [32]. Second, Lumer et al. mix different views of their layer architecture without providing sufficient guidance to the reader. They begin by describing the Vp layer as a grid of 8×8 macro-units, with two “selectivities within a macro-unit”, each containing “a collection of 5×5 topographic elements, each of which corresponded to a contiguous location in retinal space”, before proceeding to state that “[t]opographic elements in Vp were organized in maps of 40×40 elements for each of the two modeled orientation selectivities.” We find it difficult to interpret this description unambiguously. We are in particular in doubt about the localization of macro-units and topographic elements in retinal space. In our view, the most parsimonious interpretation is as follows: 5×5 topographic elements placed in each of 8×8 macro-units result in a grid of 40×40 topographic elements.” This interpretation is sketched in Fig. 3.


Towards reproducible descriptions of neuronal network models.

Nordlie E, Gewaltig MO, Plesser HE - PLoS Comput. Biol. (2009)

Interpretation of Lumer [10] model architecture.The most parsimonious interpretation of the description of the primary visual cortical area Vp given by Lumer et al, is as two layers of 40×40 topographic elements, representing horizontal and vertical orientations, respectively.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2713426&req=5

pcbi-1000456-g003: Interpretation of Lumer [10] model architecture.The most parsimonious interpretation of the description of the primary visual cortical area Vp given by Lumer et al, is as two layers of 40×40 topographic elements, representing horizontal and vertical orientations, respectively.
Mentions: The most detailed explicit model studied here is the thalamocortical model presented by Lumer et al. [10]. The description of the cortical areas in this model (Vp and Vs), while complete, lacks in our opinion the clarity desirable of a good model description, and may thus help to identify rules for ideal model descriptions. For one, discussions on model design and properties are embedded in the model description, e.g., the reduction of a total of 32 “combinations of response selectivities” to just two included in the model, and a comparison of the number of neurons in the model to that found in animals. We believe that design decisions and model review should be kept separate from the model description proper for the sake of clarity, since they are independent intellectual endeavours [32]. Second, Lumer et al. mix different views of their layer architecture without providing sufficient guidance to the reader. They begin by describing the Vp layer as a grid of 8×8 macro-units, with two “selectivities within a macro-unit”, each containing “a collection of 5×5 topographic elements, each of which corresponded to a contiguous location in retinal space”, before proceeding to state that “[t]opographic elements in Vp were organized in maps of 40×40 elements for each of the two modeled orientation selectivities.” We find it difficult to interpret this description unambiguously. We are in particular in doubt about the localization of macro-units and topographic elements in retinal space. In our view, the most parsimonious interpretation is as follows: 5×5 topographic elements placed in each of 8×8 macro-units result in a grid of 40×40 topographic elements.” This interpretation is sketched in Fig. 3.

Bottom Line: Progress in science depends on the effective exchange of ideas among scientists.This hinders the critical evaluation of network models as well as their re-use.We believe that the good model description practice proposed here, together with a number of other recent initiatives on data-, model-, and software-sharing, may lead to a deeper and more fruitful exchange of ideas among computational neuroscientists in years to come.

View Article: PubMed Central - PubMed

Affiliation: Department of Mathematical Sciences and Technology, Norwegian University of Life Sciences, Aas, Norway.

ABSTRACT
Progress in science depends on the effective exchange of ideas among scientists. New ideas can be assessed and criticized in a meaningful manner only if they are formulated precisely. This applies to simulation studies as well as to experiments and theories. But after more than 50 years of neuronal network simulations, we still lack a clear and common understanding of the role of computational models in neuroscience as well as established practices for describing network models in publications. This hinders the critical evaluation of network models as well as their re-use. We analyze here 14 research papers proposing neuronal network models of different complexity and find widely varying approaches to model descriptions, with regard to both the means of description and the ordering and placement of material. We further observe great variation in the graphical representation of networks and the notation used in equations. Based on our observations, we propose a good model description practice, composed of guidelines for the organization of publications, a checklist for model descriptions, templates for tables presenting model structure, and guidelines for diagrams of networks. The main purpose of this good practice is to trigger a debate about the communication of neuronal network models in a manner comprehensible to humans, as opposed to machine-readable model description languages. We believe that the good model description practice proposed here, together with a number of other recent initiatives on data-, model-, and software-sharing, may lead to a deeper and more fruitful exchange of ideas among computational neuroscientists in years to come. We further hope that work on standardized ways of describing--and thinking about--complex neuronal networks will lead the scientific community to a clearer understanding of high-level concepts in network dynamics, and will thus lead to deeper insights into the function of the brain.

Show MeSH
Related in: MedlinePlus