Methods for determining numbers of cells and synapses: A case for more uniform standards of review

Neuron and synapse numbers are important assays in neuroscience. These numbers are estimated by one of four methods: 1) profile counts, 2) assumption-based methods, 3) serial reconstructions, and 4) stereological methods. The criteria for these methods are diverse. This creates a disparity in that some reviewers accept estimates from any of these methods, while others accept only specific methods. An equally important issue is the diversity of sampling strategies, since unbiased estimates of neuronal or synaptic numbers are contingent upon both counting and sampling techniques. The purpose of this commentary is to institute a dialog that will lead to a better understanding of the strengths and weaknesses of the above methods, and to propose guidelines that should lead to more uniform and thus fairer judging of the studies that provide estimates of neuron or synapse numbers. In addition, adoption of more uniform standards for obtaining unbiased numerical estimates should result in the generation of an unbiased database that will be of considerable use in future studies.