I summarize recent computational and experimental work that addresses the inherent variability in the synaptic and intrinsic conductances in normal healthy brains and shows that multiple solutions (sets of parameters) can produce similar circuit performance. I then discuss a number of issues raised by this observation, such as which parameter variations arise from compensatory mechanisms and which reflect insensitivity to those particular parameters. I ask whether networks with different sets of underlying parameters can nonetheless respond reliably to neuromodulation and other global perturbations. At the computational level, I describe a paradigm shift in which it is becoming increasingly common to develop families of models that reflect the variance in the biological data that the models are intended to illuminate rather than single, highly tuned models. On the experimental side, I discuss the inherent limitations of overreliance on mean data and suggest that it is important to look for compensations and correlations among as many system parameters as possible, and between each system parameter and circuit performance. This second paradigm shift will require moving away from measurements of each system component in isolation but should reveal important previously undescribed principles in the organization of complex systems such as brains.circuit dynamics | neuronal variability | neuronal homeostasis | dynamic clamp A ll experimental biologists face a daily conundrum: on the one hand, we know that all individual biological organisms, be they lobsters, cats, or humans, are distinct individuals. On the other hand, we must do experiments on multiple individuals to ensure the reliability of our results. As biologists wishing to understand how the function of a cell, a circuit, or a brain depends on the properties of its constituent processes, we confront two issues: (i) all our data come with associated measurement error (often difficult to assess), and (ii) there is considerable natural variability in the populations we study. Consequently, we conventionally rely on statistics calculated from populations to assure ourselves that our measurements are reliable. Most commonly, we report mean data, with the underlying assumption that these means capture something akin to a "platonic ideal" of the individual neurons or animals whose properties were measured. Although this strategy has been enormously useful over the years, it has many limitations, some of which I discuss below.
Multiple Solutions and Failure of AveragingDespite the widespread use of means, computational work has shown unambiguously some of the dangers and confounds that can come from exclusive reliance on mean data (1-3). One example of this comes from a study in which several thousand model neurons were generated by randomly picking the maximal conductances of the five different ionic conductances in the model (2). Fig. 1 shows three examples of single-spike bursters chosen from a population of 164 single-spike bursters generated in this study. Alth...