Internal representation
An internal representation is an information-bearing state or encoding within a system that represents aspects of its input, environment, or task.
Machines that Think
An internal representation is an information-bearing mental or computational state that stands for aspects of the world or a task.
- How it is used
- Uses internal representation to explain how neural networks form complex responses and why fluent behavior need not imply understanding.
- How deep it goes
- Explains successive transformations and capacity, then examines representational convergence and its limits without implementation details.
- Examples
- Uses linguistic situations and independently trained language and image models to illustrate representational distinctions and convergence.
- What you can do after
- Evaluate whether fluent behavior or aligned representations warrant claims of understanding or a world model.
Read how Machines that Think explains internal representation →
Sutskever's List
A representation is an internal neural encoding of input information that can organize features relevant to a task.
- How it is used
- Uses representation complexity to interpret the transition from memorization to generalization during grokking.
- How deep it goes
- Tracks complexity rising and falling during training, linking the decline to a simpler generalizing pattern.
- Examples
- Uses grokking experiments on modular multiplication and other modular arithmetic tasks.
- What you can do after
- Understand how representation complexity can change as a network shifts from memorization to generalization.
Read how Sutskever's List explains internal representation →
Also covered in
What to know first
- language modelA language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
Related concepts
- neural networkproduced by · A neural network is a computational model composed of interconnected, parameterized units or transformations that learn to map inputs to outputs.
- behaviorismcontrasts with · Behaviorism is a psychological approach that explains behavior through observable environmental stimuli, responses, and consequences rather than internal mental processes.
- language modelused by · A language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- world modelpart of · A world model is an internal computational representation of entities, relationships, states, and dynamics in an environment that supports prediction, reasoning, or action.
- natural languageconfigured by · Natural language is a human communication system that develops through social use and is expressed in spoken or written form.
Questions about internal representation
What is internal representation?
An internal representation is an information-bearing state or encoding within a system that represents aspects of its input, environment, or task.
Which Manning books explain internal representation?
Machines that Think and Sutskever's List.
What should I know before learning internal representation?
Manning books that teach internal representation usually introduce language model first.
What is the difference between internal representation and behaviorism?
An internal representation is an information-bearing state or encoding within a system that represents aspects of its input, environment, or task. Behaviorism is a psychological approach that explains behavior through observable environmental stimuli, responses, and consequences rather than internal mental processes.
What is internal representation part of?
Internal representation is part of world model. A world model is an internal computational representation of entities, relationships, states, and dynamics in an environment that supports prediction, reasoning, or action.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.

