Representation

The term representation is used everywhere in cognitive neuroscience, often without much of a precise definition. In the computational cognitive neuroscience framework used here, we can provide a precise definition of a representation:

A representation is a pattern of neural activity distributed across a population of neurons, where the similarity structure of these patterns captures the overall semantic meaning.

Ultimately, meaning can be grounded in terms of how the rest of the brain will respond to a pattern of activity in one part of the brain. This may seem circular, but this circularity is inevitable in the sense that everything ultimately must depend on the fact that neurons and networks of neurons, interconnected with synaptic weights, are the basic mechanisms underlying brain function.

Based on our understanding of how neurons work, they integrate weighted synaptic inputs from many other neurons. This is why meaning or representation is based on the overall patterns of activity across populations of neurons: that is what drives other neurons to respond as they do.

To summarize:

Meaning is in the patterns of neural activity.

Knowledge is in the synaptic weights that shape the neural responses to these patterns of neural activity.