p.1 Psychology is the study of thinking, feeling (emotions), perceiving and acting... Cognitive science is the attempt to study cognitive phenomena in a way not unlike the way the physical sciences study material phenomena.

p.7 There are right and wrong ways of using symbols which are meaningful for us.

p.8 Our cognitive skills have their beginnings in the flow of symbolic activity of ordinary life in co-operative activities with other people, particularly in the family. Vygotsky's importance for cognitive psychology comes from his work in unraveling the complex processes by which the cognitive and practical skills of adults are acquired by infants and young children in social interactions. Higher order cognitive functions, he said, appear first in the relations between people and only later as part of an individual's mental endowment. First of all we think publicly and collectively with the assistance of others. Only later do we get the knack of thinking privately.

p.12 Throughout this course we will use the word 'grammar' for the systems of concepts and their symbolic bearers by means of which we categorize and make sense of our experiences. A grammar, then, can be expressed as an open set of malleable rules for using various symbol systems correctly.

p.12 Ludwig Wittgenstein. Grammars: rules for the correct use of symbols.

p.12-13 Scientific realists feel free to speculate in disciplined ways about the state of the world beyond the limits of perception. To do so rationally they must have in mind certain ideas about what kinds of things, properties, processes, qualities and so on the world may contain. A catalog of what is taken to be really real in some domain of enquiry is its ontology... An ontology will be among the presuppositions of a science at each moment in its development. Therefore philosophy of science must include discussions of ontology, the general assumptions about the presumed nature of the entities, structures, properties and processes both observable and unobservable characteristics of the relevant domain of enquiry.

p.14 The history of science illustrates very clearly that assessments of the ultimate value of this or that ontology may not be wise until it has been tried out in many ways and in many contexts.

p.16 Inevitably, psychology will be a hybrid science... Naturalistic studies of ordinary ways of thinking that make use of language and other symbolic systems will give us insight into the culturally and historically diverse phenomena of thinking, acting and feeling. Neurological studies will give us insights into the cerebral tools we use to accomplish the cognitive tasks contemporary life presents us with. How do we bridge the gap between naturalistic studies of meaningful actions by active people and neurological research programs studying material processes, so that the latter are relevant to the former? ...The answer is to be found in developments in artificial intelligence, with the help of which we can build effective and abstract models of the possible mechanisms of cognition, based on abstract models of processes of cognition.

p.28 Realists hold that human beings have access to the world not only through their senses, the world as perceived, but also through the use of the imagination, the world as conceived. The mature natural sciences are directed to bringing as much of the world as conceived or imagined within the reach of experiment. [JLJ - I imagine that I am a realist. No, I realize that I imagine. No. It seems that the mind is formed when things sensed and things imagined exist in a twilight together, sparking us to action via visions of things which may yet not be present and may not even come to pass, accompanied by the echoes of the social conversation which exists around us. We constantly prepare for an imagined future which is constantly changing - this is why action is never complete but always needs adaptive capacity to complete whatever emerges from the unforeseen.]

p.42 A model is a tool for thinking, one of the ways we make representations of some subject matter the better to think about it.

p.45 Taken together, abstraction (not every detail in a stratum need be reproduced in the geological model) and idealization (not every kink and break in the strata boundaries need be reproduced in the model) lead to simplification of the natural state of affairs in the model representing it.

p.46 Just how can a scientist build a model of something hitherto unknown? ...Models evolve as research programs are pursued.

p.47 A scientific model provides a resource for a certain prescription to which an object, attribute, state, substance or structure must more or less conform... By building a model we have managed to create a picture of a process that, because of its vast scale in time and space, we could never observe

p.49 Aronson (1991), Way (1992) and others have taken up the idea that model building is based on finding subtypes, within an existing type hierarchy, one of which is the source of the model and another its subject. Creating the model by abstraction and idealization of attributes from the source creates another subtype at the same level in the type hierarchy. Similarly, the subject of the model falls into place as another subtype in the same set of places in the type hierarchy... it is because they [the model and its subject] exemplify the same supertype that they are analogous... Provided we were not tempted to crystallize relations in the working hierarchy too soon into a rigid logic of necessary and sufficient conditions, it would retain its fluidity and dynamic character... Indeed, even in the most sophisticated uses of such hierarchies, the point of the supertype-subtype relation is to fix which similarities and differences should be attended to in building and assessing models. [JLJ - one could argue loosely that the game of chess has a dynamic complexity similar to the 'life-world' that we as humans live and move in. The human mind has evolved so that one might unravel the complexity of this life-world and determine how to 'go on'. Perhaps we need to model the chess-world in the same way - something like a 'mind' in order to unravel the complexity present on the gameboard, and determine how to 'go on'. That means, our model would have to do something similar to 'thinking'.]

p.50 In order that a model can stand in a representational relation to that of which it is a model, both model and subject must be members of the same type hierarchy. Their relationship - for example, that the model is an analog of its subject - will be determined by what they each inherit from the lowest common supertype they fall under in the type hierarchy.

p.50 In using analytical models to reveal the structures and processes in observable phenomena that would otherwise be too obscure, too complex or too fleeting to allow scientific work to be done on them we make use of certain standards to assess the value of the model that is being employed. The two of most importance are clarity and fruitfulness... these are... commonsensical

p.50 Clarity must not be confused with simplicity.

p.50 Fruitfulness is just the power of the analytical model to enable the user to see relationships that might have been obscured by too much detail in the original phenomenon... Analytical models have no pretensions to independent powers of representation. They are one kind of heuristic model, useful but not scientifically creative.

p.51 This opens up the possibility of a different concept of truth for scientific theories, based on the plausibility of the relevant models. We could call it 'iconic truth,' the truth of pictures as opposed to the truth of statements, verbal presentations of fact.

p.51 The second major application of the newly revived notion of a model has to do with the role and nature of experiments as a source of knowledge... Models can also be the work of engineers, laboratory technicians or instrument makers. Playing with such devices is experimenting on a model world.

p.52 What we learn from manipulating the model we can feed back into our knowledge of its subject.

p.53 If theories are taken as instructions for constructing models, then the standing of theories is just the standing of the models they can be used to create.

p.63 Once we acknowledge the role of symbolic systems, and especially language, in both the formulation and the expressions of our thoughts we can see that there is a common realm, the realm of symbols and their manipulation according to rule.

p.64 The agentive picture: the second Cognitive Revolution... In the agentive picture a person is the prime inaugurator of meaningful actions. Psychological phenomena are the products of human beings actively engaged in carrying through, or trying to carry through, their projects.

p.93 Accounting for actions in terms of the future goals towards which they are directed is called 'teleological explanation'.

p.103 J.S. Bruner (1973, 1983)... His experiments suggested that there must be prior cognitive schemas that are involved in perception and utilized in cognitive processes of which we are unaware.

p.104 we must have

Observable stimulus (retinal sensation) together with unobservable Cognitive process ('knowledge utilization') → Observable response (recognition of word)

p.105 Let us begin with a reminder as to what the realist shape for a [cognitive] science involves. The main thrust of this interpretation of science, particularly the development of hypotheses about and models of hypothetical generative mechanisms is that, though unobserved by investigators, such mechanisms are necessary for the production of the observed phenomena.

p.106 There is a long history of attempts to make machines that will do what people can do.

p.110 is it true that every cognitive process can be represented by a computable function?

p.112 Our aim is to model thinking.

p.118 Human cognitive practices depend on our ability to recognize the intentionality of signs.

p.118 the 'Chinese room'... John Searle (1980)... thought experiment was designed to show that there is no place for the use of mentalistic language in describing the performance of computing machines when running programs... To perform cognitive operations a material system need only operate according to certain rules on meaningless symbols, material states of that system.

p.122 Should we call what a computer does 'thinking'? ...neither computers nor brains can properly be said to think. However, a brain is part of a person, who can think... it is the persons who think!

[JLJ - interesting, but I can construct a sign-symbol system, which organizes and prioritizes reflex-formed musings of how to go on, into plausible sequences and projects, and constructs scenarios which compete for attention. The output of this device (how to 'go on') can be somewhat intelligent action. We as programmers simply animate the machine in future situations which we mostly understand to be useful (such as retrieving data from a database), and we let our users tell us when the behavior is not useful to them. We fix those situations.

The machine might not be 'thinking', but what it will be doing, internally, will be the equivalent to asking itself questions, and forming answers, in a pretty close representation to an 'internal conversation'. When the machine is not sure what to do, it pops up a dialog box and asks the user a question. It then gets clarification of how to go on. The intention does not come from the machine, it comes from the user.

Thinking cannot properly be focused unless it involves moving a person, a physical body, through a complicated and ambiguous world, full of social signs and symbols, trolling for opportunities and managing threats, preparing and executing projects, etc.]

p.122 Cognition is the management of meanings... What are the necessary conditions for a mark to be meaningful? ...Intentionality is a mark of real cognition... Intentionality is a property not of a single isolated sign but of a sign in a well established context of practices with which people accomplish cognitive tasks.

p.125 Rules do not determine what happens in the future. They determine what should happen... If a rule is represented by an instruction in a program then it is transformed ontologically. It is no longer a norm but a cause... Only people conform to or violate norms. Brains, and computing machines, do what they do.

p.126 Rules... guide action but do not fix what is to happen.

p.127 In practice, computer programmers construct restricted frames, as abstract representations of a cognitive system... No frame can be prepared in advance for every contingency.

p.137 Using the available symbolic resources, people undertake cognitive tasks. The concept of 'skill' can be used to link individuals with the matrix of interpersonal symbolic and practical interactions.

p.141 The overall project of creating an adequate cognitive psychology is built on the principle that when people are thinking they are actively engaged in carrying out cognitive tasks according to local standards of good work. The generic model or source of all forms of cognition is the cognitive task performed with symbolic tools.

p.141 Thinking... deciding and so on not only often take the form of conversations, but when carried out in other symbolic media than language the cognitive processes are nevertheless conversation-like. They are structured by meaning relations and can be seen to aspire to normative standards of correctness and propriety.

p.142 A sign is a meaningful symbol as far as people use it for performing cognitive tasks. Cognitive tasks are those jobs that people work at by using signs as meaningful symbols. Tasks and the symbolic tools by which they are accomplished mutually define one another... The task/tool circle is true of every human practice... Having procedural knowledge of the practice, how to do it whatever it is, brings along with it an understanding of the uses to which the relevant means are to be put.

p.147 According to the 'discursive' point of view... psychology is primarily the study of processes - streams of human actions and interactions. These can be understood in terms of their meanings for the actors and interactors and the norms and traditions that are generally accepted. Many of these streams of meaningful actions make their sense in terms of narratives, story lines well known in the culture. Within this general scheme conversation is the most useful... model for analysing such streams of action. Adopting this model... invites the researchers to treat all that people do... as if it were a kind of conversation... Cognitive psychology is the study of how individuals come to bring about meaningful actions in contexts that are themselves meaningful.

p.147-148 The whole of human life is enmeshed in loosely linked and fuzzy-edged standards of correctness and propriety. For the most part such standards are presupposed in the way we think and act... They can be expressed for scientific purposes in terms of systems and clusters of rules. We have adopted Wittgenstein's term 'grammar' to refer to useful ways of grouping such sets of rules into semi-coherent systems.

p.150 Dennett... asks what it is that organizes our capacity to be so good at understanding other people. The answer lies in the fact that we can predict each other's behavior. We can do so because we adopt 'the intentional stance' to one another. To adopt this stance is to treat 'the object whose behavior you want to predict as a rational agent with beliefs and desires... exhibiting what Bretano called intentionality' (Dennett, 1987: 15).

p.151 according to the discursive point of view, thinking and acting intentionally is not a manifestation of a mentality. It is mentality. The intentional stance is what we each take to ourselves as well as to others.

p.156 According to Vygotsky (1962), every higher cognitive function exists twice over, once in the social environment of a developing human being and then as a competence or cognitive skill to be exercised by that being. The mediation between social environment and individual person is achieved by a kind of psychological symbiosis. An infant's attempts at skilled performance, be they motor or intellectual, are supplemented by someone more skilled.

p.158 The trick upon which the possibility of a unified cognitive science depends is to shift the focus from entities to discourses.

p.159 A certain electronic device is a 'calculator' only in relation to the task it is used to perform.

p.169 According to discursive psychologists, cognitive tasks are accomplished by the skillful use of symbols. The conversational model directs our attention particularly to linguistic performance... Cognitive projects... are often accomplished linguistically, or by symbolic activities that are language-like... Scientific thinking often involves thought experiments, performed in the mind with imaginary apparatus. Many technical cognitive tasks are done by the use of graphs and diagrams. These symbolic devices are first order tools. People use them to accomplish their cognitive projects.

p.181 Taking the conversation model seriously would suggest that 'grammar' may not only serve as a metaphor for a wide range of ordering principles but also have a literal role in the psychologically relevant analysis of what people do.

p.198 In Chapter 11 we will encounter the phenomenon of knowing how to do something when it is not presupposed that a knowledgeable and skillful person could recite the relevant rules explicitly. This is called 'procedural memory'.

p.216 The task of cognitive psychology will be to create working models of the cognitive processes by which patterns of meaningful actions... are produced... Some cognitive processes are private and individual. The Vygotskian principle tells us that they are derived from public and collective conversational performances... Thinking is a symbolic activity, whether it is performed privately or publicly. [JLJ - perhaps by a computer or a person]

p.217 How can we build the necessary bridge from our knowledge of cognitive tasks to hypotheses about the tools by which we perform them? ...Artificial intelligence projects, properly interpreted, are exemplars of how psychology as 'the hybrid science' should be pursued. Programs represent sets of rules for the performance of cognitive or material tasks, extracted from an analysis of acceptable practice. The computational system represents the material groundings for the dispositions, skills and so on implied by the ability of the person to perform the cognitive and material tasks correctly. That can serve as a model for well grounded hypotheses about how the brain and nervous system work as tools for these and other tasks.

p.217 For the most part the processes by which people carry out cognitive tasks are unobservable. They are not available to conscious inspection, or to neurological research... a science of psychology should open up by the development of powerful and plausible models of the unknown and unobservable processes going on in the second order tools by which people bring about the phenomena in question.

p.227 Modeling, the heart of cognitive science, as it is of any scientific enterprise, is constrained by two essential relations. It is constrained by what we know about public conduct and procedures and skilled performance on the one hand, and by neurological possibilities on the other.

p.255 Everyday activities depend on the human ability to classify and categorize... Could the diagnostic techniques developed in setting up diagnostic programs for physicians, for example, be hi-jacked as the basis of a psychology of the human ability to classify butterflies and/or to recognize the grammatical categories of words?

p.258 According to Sowa (2000), five desiderata [JLJ - desired things] should be met by any system that can be used to represent a body of knowledge.

1. There must be a symbol system that maps on to real-world things.
2. Such a system has ontological commitments. [JLJ - an ontological commitment is said to be necessary in order to make a proposition in which the existence of one thing is presupposed or implied by asserting the existence of another.]
3. A knowledge representational system must include a 'theory', which describes the ways the classified things or substances behave.
4. There must be a medium of efficient computation.
5. There must be a medium for human expression.

p.259-260 How do we distinguish essential properties from the propria, those properties which are universal but not used as part of the essence? ...Generally, a real essence is derived from the relevant theory and supported by complicated patterns of reasoning in the management of... models... Real essences explain why the set or type or species from a stable cluster

p.264 Frederick Waismann, philosopher and sometime confidant and amauensis [JLJ - secretary or assistant] of Ludwig Wittgenstein, was struck by the fact that real classification systems, such as the periodic table of the chemical elements, were quite capable of absorbing new items of knowledge without generating fatal internal contradictions (Waismann, 1968: 41-3, 95-7).

p.278 Conversation is both an example of a cognitive medium and an analytical model for analysing performances in non-linguistic cognitive media.

p.285 A project is scientific only as far as a well established system of classificatory concepts, a taxonomy is available, and there is a working type hierarchy for constructing plausible models of unobservable entities, properties and processes. Ideally, these essential features of a scientific approach to any domain are linked into a coherent overall pattern.

p.289 The main exemplar for discursive psychology is the conversation in which two or more people carry out some cognitive task in the course of speaking (or sometimes writing) to one another. Real conversations are exceedingly complex phenomena... realizing ever shifting personal intentions, consensual agreements and patterns of mutual positioning with respect to the right to speak and the obligation to listen and/or respond. In the course of conversing people create, maintain, transform and abrogate social relations. In the course of conversing people adjudicate disputes, [and] arrive at decisions