Abstract:figure legend Neural networks execute the transformation of inputs into the neural activity that enables the execution of a task. Thus, they have the potential to establish causal relationships between the input, the elements of a circuit, the network activity, and eventually behavior. When the purpose of building a model is to understand the mechanisms that enable behavior in biological circuits, the design of network models should be guided by three principles: (1) biological constraints are necessary (e.g., connectivity), (2) the network activity should match the recorded activity, and (3) the network should be able to perform the task of interest. Machine‐learning‐based training can help accomplish these goals. Due to the staggering complexity of the brain and its neural circuitry, neuroscientists rely on the analysis of mathematical models to elucidate its function. From Hodgkin and Huxley's detailed description of the action potential in 1952 to today, new theories and increasing computational power have opened up novel avenues to study how neural circuits implement the computations that underlie behavior. Computational neuroscientists have developed many models of neural circuits that differ in complexity, biological realism, or emergent network properties. With recent advances in experimental techniques for detailed anatomical reconstructions of large‐scale activity recordings, precise biological data has become more available. The challenge when building network models is to reflect experimental results, either through a high level of detail or by finding an appropriate level of abstraction. Meanwhile, machine learning has facilitated the development of artificial neural networks (ANN), which are trained to perform specific tasks. While they have proven successful at achieving task‐oriented behavior, they are often abstract constructs that differ in many features from the physiology of brain circuits. Thus, it is unclear whether the mechanisms underlying computation in biological circuits can be investigated by analyzing artificial networks that accomplish the same function but differ in their mechanisms. Here, we argue that building biologically realistic network models is crucial to establishing causal relationships between neurons, synapses, circuits, and behavior. More specifically, we advocate for network models that consider the connectivity structure and the recorded activity dynamics while evaluating task performance.

Recurrent neural circuits overcome partial inactivation by compensation and re-learning

Dis-inhibitory neuronal circuits can control the sign of synaptic plasticity

Interactions between circuit architecture and plasticity in a closed-loop cerebellar system

Partial observation can induce mechanistic mismatches in data-constrained models of neural dynamics

Deep inverse modeling reveals dynamic-dependent invariances in neural circuit mechanisms

Biologically informed cortical models predict optogenetic perturbations

The sloppy relationship between neural circuit structure and function

Neural Circuit Inference from Function to Structure

A Neural Circuit Model of Flexible Sensorimotor Mapping: Learning and Forgetting on Multiple Timescales

Explaining neuroplasticity using equivalent circuits: A mechanistic perspective

Neural circuits as computational dynamical systems

A Possible Neural Circuit for Decision Making and Its Learning Process.

Downstream network transformations dissociate neural activity from causal functional contributions

The brain as an efficient and robust adaptive learner

Exploring the Architectural Biases of the Canonical Cortical Microcircuit

No neuron is an island: Homeostatic plasticity and over-constraint in a neural circuit

One Step Back, Two Steps Forward: Interference and Learning in Recurrent Neural Networks

Top-down modulation in canonical cortical circuits with short-term plasticity

How to incorporate biological insights into network models and why it matters

Recurrent activity within microcircuits of macaque dorsolateral prefrontal cortex tracks cognitive flexibility

Using local plasticity rules to train recurrent neural networks