Summer 2026
From biological brains to engineered systems
About the Course
Neural computation spans the full arc from biological brains to engineered systems. This course begins with the anatomy and physiology of the human brain, traces the developmental program that wires 86 billion neurons, formalizes the plasticity rules that enable learning, and examines theories of consciousness. The second half turns outward: connectome-to-function mapping, whole-brain emulation, biological computing with living neurons, brain–computer interfaces, and the specialization of silicon hardware from CPUs through GPUs to model-specific ASICs.
Prerequisites
Linear algebra and probability. Recommended: machine learning.
Textbook
Iddo Drori, The Brain, Forthcoming, 2026.
Instructor
Prof. Iddo Drori
Course Schedule
Course overview; scope and themes; organization of the brain sciences; the computational lens on neuroscience
Neurons, glia, action potentials, synaptic transmission, neurotransmitters, brain regions
Cortical circuits, sensory and motor systems, computational organization, brain–AI parallels
Morphogen gradients, cell fate specification, axon guidance, spontaneous activity, glial sculpting
Critical periods, whole-brain cell atlases, genomic foundation models
Hebbian learning, LTP/LTD, STDP, dendritic and structural plasticity
Three-factor neuromodulated learning rules, sleep replay, memory consolidation, plasticity–stability dilemma
The self-scanning hypothesis, Global Neuronal Workspace theory, Integrated Information Theory, higher-order and recurrent processing theories
Adversarial testing (COGITATE), cross-species evidence, theory-derived indicators for machine consciousness, substrate independence
Brain-wide activity maps, structure–function connectomics, neural foundation models, digital twins of the brain
Reading a connectome, simulating neural dynamics, coupling to a virtual body, the virtual fly
Neurons on electrode arrays, closed-loop active-inference framework, organoid intelligence
Modalities from EEG to intracortical arrays, decoding algorithms, speech neuroprosthetics
CPUs to GPUs, AI accelerators, transformer-specific ASICs, model-specific chips; course synthesis
Hands-On Labs
Lab 01
Simulate Hodgkin–Huxley and leaky integrate-and-fire neurons, build chemical and short-term-plasticity synapses, decode population activity vectors.
Lab 02
Grow morphogen concentration gradients, generate Turing patterns via reaction–diffusion, model axon guidance, and simulate critical-period plasticity.
Lab 03
Implement Hebbian, Oja, STDP, BCM, and three-factor learning rules. Explore replay buffers and homeostatic synaptic scaling.
Lab 04
Build a Global Neuronal Workspace with ignition dynamics, compute integrated information Φ, model thalamocortical loops and binocular rivalry.
Lab 05
Analyze connectome graphs with degree distributions, clustering, rich-club coefficients, spectral methods, and winner-take-all network dynamics.
Lab 06
Emulate C. elegans circuits, build sensorimotor agents, and train embodied controllers with reward-modulated STDP.
Lab 07
Simulate multi-electrode array recordings, build an LIF reservoir computer, implement active inference, and play Pong with a neural controller.
Lab 08
Decode movement with population vectors, Kalman filters, and LSTM networks. Analyze channel capacity and information transfer rates.
Lab 09
Profile roofline models for CPUs, GPUs, and TPUs. Design neuromorphic crossbar arrays and run neural architecture search on the Pareto frontier.
Assessment
Problem sets combining analytical derivations, programming exercises, and critical reading of primary sources.
Research paper presentations.
In-class quizzes, cumulative, covering the textbook material across the semester.
Active engagement in class discussions and peer paper presentations.
Course Policies
All submitted work must be your own, in accordance with the Yeshiva University Code of Academic Conduct. Collaboration on homework is encouraged for discussion, but each student must write up solutions independently. Use of AI assistants must be disclosed.
Students needing accommodations should contact the Office of Disability Services and provide documentation to the instructor during the first two weeks of the semester.
Each student has 5 late days for homeworks, with at most 2 late days per homework assignment.