Event
Logik trifft auf Nervenzellen
In 1943 Warren McCulloch and Walter Pitts published a mathematical model of networks of idealized nerve cells. A unit is either active or inactive; in their initial assumptions, activation depends on excitatory and inhibitory inputs. They investigated how such networks can be described using propositional logic. Original paper
Intuitively, under suitable assumptions such a unit can behave like a logical condition: it becomes active when enough inputs are active and no inhibitory input blocks it. This was a theoretical model and a later reference point for research on artificial neural networks; it was neither a trained modern AI system nor a complete model of biological brains. The authors discussed learning and themselves emphasized limits of formal equivalence. Original paper
Sources
- Originalaufsatz und Verlagsdatensatz: Annahmen, logische Beschreibung und Grenzen des Modells.