Jerome A.Feldman
From Molecule to Metaphor
A Neural Theory of Language
Mit Press 2008

Feldman 95
Embodied Concepts

The first seven chapters summarised our magnificent neural machinery, how it develops, and how it can be studied as an information processing system. Almost all of that discussion applies to animals in general and there is much more to be learned by studying animals as information processing systems, adapting to their environment and goals. But this book is about one special adaptation, language, that is unique to humans. Human conceptual systems are inextriably linked to language.

96 The basis for concepts is categorisation. Categorisation occurs whenever a lot of data boiled down to a few all values. This happens in the retina and everywhere else in the brain, whenever a number of neurons signal to another neuron. Categorisation is not just a function of language. All living systems categorise.

Some philosophical traditions ask us to rise above our human categorisations and see the world as it really is, assuming some basic structure of nature that is independent of people. However, this is impossible for neural beings who evolved to do best-fit matching of input to the current context and goals. We have good reason to believe that there is a real physical world, but not that there is a privileged way of categorising it. People evolved to develop categories that match their situation and needs. These must be consistent with the facts about the physical and social environment or they wouldn't be of any use.

Besides the simple categories linked directly to perception and action, there are also more complex conceptual categories. The major concern in this book is with how people connect low-level information at the neural level with higher-level conceptual categories such as house, ugly, ask, truth. Walking down the street, we categorise the pavement versus the street, things that move versus things that stand still, things to step on versus things not to step on, people you know versus people you don't know, dangerous versus nondangerous things. How can a neural system form conceptual categories? To answer this question, we need to know more about conceptual categories.

Categories with prototypical members
Typical case prototypes
Ideal case prototypes
Radial categories

Basic level categories:
100 basic level categories: how do basic level categories arise?
101 Eleanor Rosch found that this categorisation happens with ordinary objects around us. If you look at a hierarchy of categories, shutters furniture > chair > rocking chair , the middle of the hierarchy is a basic level category.

In general, basic level categories have mental images associated with them. Chairs evoke an image, but generalised furniture does not. We have motor programmes for interacting with the things. For instance, we have standard motor programmes for interacting with the chair, but there's no motor programme for interacting in general with furniture. In addition, much of our knowledge is organised at the basic level. We know a great deal about how to interact with chairs and tables, but little about furniture in general. Aside from a few very general categories such as animals and vehicules, children learn basic level distinctions first.

Let us look more carefully at words and how we learn them. A central question is the relationship between words and the world. The traditional view was that the world determines which concepts are needed and words are arbitrary labels different languages use for a fixed set of concepts that the world provides. But common sense tells us that our concepts depend on how we interact with the world. This can differ widely among cultures and professions. Nevertheless, all people share the same underlying physiology and we know that many concepts and their words are determined directly by their embodiment. For example, think of the short words in English - hand, hear, hit, hot, hungry, happy.

101 The interaction of people with their physical and social environment defines various semantic spaces, such as the space of colours, emotions, or dance steps. Languages differ in how they talk about each of these semantic spaces, but all languages must have ways of expressing the conceptual primitives that all people share.

105 People are generally comfortable with the idea that words are concepts and the connections among them are entities in the mind. It also seems reasonable to associate each mental concepts of some neural structure and imagined conceptual links being captured as active neural connections… we can explain various kinds of simple mental functions, such as priming or reminding, as direct consequences of spreading activation at the neural level. That is, mental connections are active neural connections. But there is a lot more to language and thought than simple spreading activation. We need to spell out how these complex mental functions can also be realised as active neural connections.

105 Even to begin to explain the intricate processes of language learning and use requires a way of describing language and thought processes.
A long and distinguished tradition has tried to define some formal „
laws of thought“ characterising meaning and reasoning. Attempts to define exact grammatical rules for language go back to a Sanskrit scholars of many centuries ago. The current work of many linguists is concerned with trying to describe the form and meaning of language in strict mathematical formalisms, deliberately avoiding any connection to human bodies or experience. Another group, the cognitive linguists, studies how language interacts with other mental functions but they have lacked formal notations for expressing their insights.

Recent developments have suggested the possibility of finding a means of scientific expression and rich enough to express the links of language to embodied cognition and also sufficiently rigorous to support simulation and direct experimental testing.The scientific notation that I adopt here is neuronally inspired and based on the abstract model of neural computation.

106 Even if we have a good way of describing the complexities of grammar, how could we explain language use in terms of neural structures and activity? The key insight is that, for many purposes, the brain can be viewed as an information processing system. All of the brains intricate circuitry and exquisite processes of development and learning evolved to enable people to extract information from the environment and use it to achieve their goals.

Chapter 2 presented a general discussion of the information processing perspective on the functioning of animals. I will now explain some of the detail the mechanisms for describing neural computation that cognitive scientists have developed. The goal of their effort has been to establish the precise scientific methodology for specifying how the functioning of neural structures supports various behaviours. The theory of neural information processing is just what we need to build a bridge between brain and mind. Any such theory will need to provide a neural account of three crucial information processing functions:
1. How words and concepts represented in the brain?
2. How do these representations co-operate in mental activity?
3. How does the brain learn language?
There are no definitive answers to any of these questions, but enough is known to seriously constrain theories of language and thought. Notice that even the framing of question 2 assumes that any answers must explicitly involve a massive parallel character of our neural architecture.

Sigmund Freud
Wiliam James
Warren McCulloch, Walter Pitts
Frank Rosenblatt

Connectionist models:
Spatial and other maps:

112 Neural representation is also characterised by dozens of systematic maps - collections of linked neurons with related functionality… Neural maps also play an important role in theories and models of brain function.

Cognitive Linguistics