Efekan Salman | Computational Neuroscience & Bio-Inspired AI

INTERACTIVE FIELD NOTE / 01

Make a neuron fire. Change a synapse.

Two small experiments show how input current produces spikes and how spike order changes a connection.

01 / LIFTeaching model

When does a neuron fire?

Increase the input current. Voltage rises to a threshold, then the neuron spikes and resets.

Input current2.2 nA

0 nA3 nA

Firing threshold−55 mV

−60 mV−45 mV

MEMBRANE VOLTAGE0 spikes / 200 ms

Euler simulation: τ = 20 ms, R = 10 MΩ, rest/reset = −70 mV, refractory period = 4 ms.

02 / STDPTeaching model

Which spike comes first?

Move the postsynaptic spike. Pre before post strengthens this example synapse; reversing the order weakens it.

Post − pre timing+12 ms

−40 ms0 ms+40 ms

PresynapticPostsynaptic

EXAMPLE WEIGHT CHANGE+0.110Connection strengthens

Illustrative pair STDP: A+ = 0.20, A− = 0.22, τ = 20 ms. One pair; no network training or measured result.

Background reading: Neuronal Dynamics · Bi & Poo (1998)

01 // Background

About Me

I'm Efekan, a developer and computational neuroscience explorer focused on simulating artificial consciousness, neural decision-making, and biologically inspired systems through code.

Transitioning from backend systems to modeling neural mechanisms, I build biophysical spiking neural networks, transformer-brain alignment probes, and cognitive architectures simulating internal motivational drives.

AUC 0.963

Neuroanatomical decoding accuracy on Allen Brain Atlas (AxonLM)

LIF & SNN

Biophysical Spiking Neural Network & STDP synaptic plasticity

02 // Further Work

Other Projects

Multi-Agent ABM

ShipOfFools

AI-driven social agent-based modeling (ABM) exploring human psychology, power dynamics, trust network formation, and dynamic anomaly detection.

Neural Decoding

BrainDecode & Neural Dynamics

Decoding brain connectivity, cognitive confusion states, and EEG neural time-series data using Brian2 and signal processing pipelines.

Production MLOps

Production ML Systems

End-to-end production ML pipeline & Dockerized FastAPI service engineered with strict pre-commit validation, Pydantic data contracts, and predictive modeling.