Phenomenological models of the brain are simplified models that follow a top-down strategy: the model's focus is on the function performed and not in the details of the structure that performs the task. This latter type of approach would be typical of so-called realistic models that use a bottom-up approach. Phenomenological models try to capture the minimum characteristics capable of simulating the basic operations of the modelled phenomenon. Examples of these characteristics in the case of neuron function are the different types of neural plasticity (synaptic and non-synaptic), the various activation functions (Heaviside, linear, sigmoidal, gaussian), and the different kinds of inhibition (all or nothing, subtractive and shunting inhibition). These models ignore the details that could make a model computationally inefficient, allowing a faster calculation on ordinary computers. For this reason, they can be embedded in devices for practical purposes where calculation speed, transportability, and economics in implementation are decisive like in microcontrollers, application-specific integrated circuit (ASIC), or field-programmable gate arrays (FPGA).

When selecting variables and parameters, phenomenological models use the parsimony principle, selecting only those variables that are relevant at a functional level and ignoring the parameters and variables used to describe the type of substrate used (in the case of the brain, a carbon-based biological substrate). Because they are more substrate-independent, and due to their computational efficiency, phenomenological models easily allow their transfer to hybrid platforms. These platforms use machine learning techniques, or artificial intelligence applied to tasks such as pattern recognition, process control, preventive maintenance, time series prediction, etc. Due to their hybrid characteristics, phenomenological models have enormous potential in front-end interfaces between computers and the nervous system, such as in brain-machine systems, neuroprosthetics, and optogenetic brain interfaces.

In this Special Issue, we welcome various types of contributions involving phenomenological models of the brain and nervous system, from theoretical mathematical models to practical implementations. We are especially interested in models applied to understand brain dysfunctions. It is easier to understand a dysfunction by looking in a function or algorithm than by digging among hundreds or thousands of variables and parameters. For this reason, phenomenological models are advantageous to understand brain diseases like Alzheimer's disease, schizophrenia, bipolar disorder, autism, hyperactivity, psychosis, phobias, etc. Original research and review articles are welcome.

Potential topics include but are not limited to the following:

• Algorithmic and mathematical models of cerebral structures: thalamus, cortex, koniocortex, cerebellum, hippocampus, basal ganglia, etc.
• Phenomenological models of brain operations: sensory and pain processing, learning, reasoning, heuristics, consciousness, etc.
• Brain disease models: Alzheimer's disease models, schizophrenia model, bipolar disorder model, etc.
• Phenomenological models embedded in microcontrollers, application-specific integrated circuit (ASIC), or field-programmable gate arrays (FPGA)
• Phenomenological models in neuroprosthetics, brain-machine and optogenetic interfaces