Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Review
, 365 (1551), 2315-28

Neuronal Network Analyses: Premises, Promises and Uncertainties

Affiliations
Review

Neuronal Network Analyses: Premises, Promises and Uncertainties

David Parker. Philos Trans R Soc Lond B Biol Sci.

Abstract

Neuronal networks assemble the cellular components needed for sensory, motor and cognitive functions. Any rational intervention in the nervous system will thus require an understanding of network function. Obtaining this understanding is widely considered to be one of the major tasks facing neuroscience today. Network analyses have been performed for some years in relatively simple systems. In addition to the direct insights these systems have provided, they also illustrate some of the difficulties of understanding network function. Nevertheless, in more complex systems (including human), claims are made that the cellular bases of behaviour are, or will shortly be, understood. While the discussion is necessarily limited, this issue will examine these claims and highlight some traditional and novel aspects of network analyses and their difficulties. This introduction discusses the criteria that need to be satisfied for network understanding, and how they relate to traditional and novel approaches being applied to addressing network function.

Figures

Figure 1.
Figure 1.
The ‘experimentally defined’ lamprey locomotor network organization (from Grillner 2003; Grillner et al. 2005; Grillner & Jessell 2009), as an illustration of a network scheme based on assumption rather than direct characterization. For example, in this scheme the I cells are not defined; are they the large crossed caudal (CC) or small crossing inhibitory neurons (ScIN; Parker 2006a)? When CC interneurons were considered as the only crossing neurons they were defined as such, but as these neurons were not consistent with segmental reciprocal inhibition (see Parker 2006a for details) two classes of undefined I neurons (which presumably represent the CC and ScINs) were added to the network scheme. This lack of definition is clearly not sufficient for a characterized network. The excitatory input (E on this diagram) to these cells is also problematic. Repeated claims, most recently by Grillner & Jessell (2009), state that they ‘excite all types of spinal neurons’, but there is no demonstrated connection between the E and the ScINs, making the claim misleading. Grillner & Jessell go on to say that ‘in lamprey and tadpole intrinsic synaptic excitation within pools of the excitatory premotor interneurons … account for the burst generation in combination with their membrane properties’, and ‘gap junctional connectivity has been reported in this interneuron pool’. Neither is established for lamprey: although it seems likely that the evidence for direct connectivity between the EINs (Parker & Grillner 2000) will support bursting, these interactions are very poorly understood, and there is no evidence for gap junctional connectivity. Detailed information is, however, available on excitatory network interneurons is available, however, from a series of detailed analyses in the tadpole (Li et al. 2006; Roberts et al. 2008, in press). For crossing inputs in the lamprey network the situation is worse: claims that they ‘cross the midline to inhibit all neuron types on the contralateral side’ (Grillner & Jessell 2009) are not justified. No CC interneuron inputs to E cells are known, and ScIN connectivity is essentially unknown (they are only known to inhibit motor neurons), despite claims that they have been determined experimentally (Grillner et al. 2005). In reality there are significant gaps in our knowledge simply at the level of the organization of this network, with obvious implications for any attempt at a functional explanation of how the network output is generated. The diagram also omits ipsilateral inhibitory inputs and crossing excitatory inputs (see Parker 2006a for a more detailed discussion). The apparent characterization here is achieved by assumption and extrapolation and by omitting reference to highlighted gaps in our knowledge (Rovainen 1983; Buchanan 1999; Parker 2006a).

Similar articles

See all similar articles

Cited by 13 PubMed Central articles

See all "Cited by" articles

LinkOut - more resources

Feedback