In the pandemonium model of perception, how is perception produced?

• A. A single detector directly matches input to a stored template.
• B. Many simple detectors (demons) compete and their outputs are combined by higher-level demons to recognize the object.
• C. A neural network learns by backpropagation to classify inputs.
• D. Perception relies solely on memory recall without sensory input.

Answer: (B)

Perception in the pandemonium model comes from a competitive cascade of simple detectors whose outputs are combined by higher-level detectors to produce the recognition. In this approach, input is analyzed by many basic detectors (the “demons”) that respond to simple features. These feature signals feed additional demons at higher levels that respond to more complex patterns, and a final decision demon weighs the competing signals to declare which object is present. This bottom-up, parallel competition and aggregation interpret the sensory data without relying on a single template match, backpropagation learning, or memory recall alone.