* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Lecture 17
Survey
Document related concepts
Agent-based model in biology wikipedia , lookup
Behaviorism wikipedia , lookup
Speech-generating device wikipedia , lookup
Perceptual control theory wikipedia , lookup
Knowledge representation and reasoning wikipedia , lookup
Existential risk from artificial general intelligence wikipedia , lookup
Personal knowledge base wikipedia , lookup
Soar (cognitive architecture) wikipedia , lookup
Neural modeling fields wikipedia , lookup
Human-Computer Interaction Institute wikipedia , lookup
Mathematical model wikipedia , lookup
Philosophy of artificial intelligence wikipedia , lookup
Human–computer interaction wikipedia , lookup
Ecological interface design wikipedia , lookup
Transcript
Cognitive models Today we will cover goal and task hierarchies Linguistic physical and device architectural What is cognitive science? Cognitive science is the science of mind and behavior. Cognitive Science Theme “Cognitive” Of or pertaining to cognition, or to the action or process of knowing. Understanding knowledge acquisition and use is the key to understanding the mind. So what is cognitive science? Cognitive science is a scientific study of the mind with special emphasis on the use and acquisition of knowledge and information. Implications An inter-disciplinary approach – Many scientific disciplines contribute to cognitive science. A computational approach – Explain information processing in terms of neural computations. Definition 1 "the study of intelligence and intelligent systems, with particular reference to intelligent behavior as computation" (Simon & Kaplan, 1989) Simon, H. A. & C. A. Kaplan, "Foundations of cognitive science", in Posner, M.I. (ed.) 1989, Foundations of Cognitive Science, MIT Press, Cambridge MA. Definition 2 Cognitive science refers to the interdisciplinary study of the acquisition and use of knowledge. It includes as contributing disciplines: artificial intelligence, psychology, linguistics, philosophy, anthropology, neuroscience, and education. Cognitive science grew out of three developments: the invention of computers and the attempts to design programs that could do the kinds of tasks that humans do; the development of information processing psychology where the goal was to specify the internal processing involved in perception, language, memory, and thought; and the development of the theory of generative grammar and related offshoots in linguistics Definition 2 (con.) Cognitive science was a synthesis concerned with the kinds of knowledge that underlie human cognition, the details of human cognitive processing, and the computational modeling of those processes. There are five major topic areas in cognitive science: knowledge representation, language, learning, thinking, and perception. Eysenck, M.W. ed. (1990). The Blackwell Dictionary of Cognitive Psychology. Cambridge, Massachusetts: Basil Blackwell Ltd. Definition 3 Generally stated, this is the study of intelligence and intelligence systems. It is a relatively new science that combines knowledge gained from a number of disciplines. These include: computer science, neuroscience, cognitive psychology, philosophy, and linguistics. As a result of the collaborative effort between these disciplines, there have been, and will continue to be, huge advancements in our understanding of human cognition. Definition in wiki Cognitive science may be broadly defined as the multidisciplinary study of mind and behavior. It draws on multiple empirical disciplines, including psychology, philosophy, neuroscience, linguistics, anthropology, computer science, sociology and biology. Definition in Plato encyclopedia Cognitive science is the interdisciplinary study of mind and intelligence, embracing philosophy, psychology, artificial intelligence, neuroscience, linguistics, and anthropology. Cognition Cognition – from Latin base cognitio – “know together” The collection of mental processes and activities used in perceiving, learning, remembering, thinking, and understanding, and the act of using those processes Information processing everywhere Perception acquiring real-time information about the surrounding environment. Language use making use of information about syntax, semantics and phonology. Reasoning combining different sources of information, deriving new information, testing consistency of information, etc. Action making use of information in action planning and guidance. Memory storing and retrieving information Cognitive Model for HCI We need to model some aspect of the user’s understanding, knowledge, intentions or processing. The level of representation differs from technique to technique – from models of high-level goals and the results of problem-solving activities, to descriptions of motor-level activity, such as keystrokes and mouse clicks. Cognitive models They model aspects of user: understanding knowledge intentions processing Common categorisation: Competence vs. Performance Computational flavour No clear divide Competence Vs Performance “Competence models tend to be ones that can predict legal behavior sequences but generally do this without reference to whether they could actually be executed by users. In contrast, performance models not only describe what the necessary behavior sequences are but usually describe both what the user needs to know and how this is employed in actual task execution.” Competence Vs Performance Competence models, therefore, represent the kinds of behavior expected of a user, but they provide little help in analyzing that behavior to determine its demands on the user. Performance models provide analytical power mainly by focusing on routine behavior in very limited applications. Classification of Cognitive Models Cognitive models for HCI are mainly classified into, hierarchical representation of the user’s task and goal structure linguistic and grammatical models physical and device-level models. Goal and task hierarchies Solve goals by solving subgoals - Mental processing as “divide-and-conquer” Example: sales report produce report gather data . find book names . . do keywords search of names database . . . … further sub-goals . . sift through names and abstracts by hand . . . … further sub-goals . search sales database - further sub-goals layout tables and histograms - further sub-goals write description - further sub-goals goals vs. tasks goals – intentions what you would like to be true tasks – actions how to achieve it GOMS – goals are internal HTA – actions external – tasks are abstractions Issues for goal hierarchies Granularity Where do we start? Where do we stop – how far to subdivide? Get down to a routine learned behavior, not problem solving Conflict More than one way to achieve a goal Techniques Cognitive Complexity Theory (CCT) Treatment of error Goals, Operators, Methods and Selection (GOMS) Hierarchical Task Analysis (HTA) -(Covered in Later Lectures) GOMS: Overview Formal representation of routine cognitive skill. A description of knowledge required by an expert user to perform a specific task. Provides a description of what the user must learn. GOMS: Classification Provides a predictive, descriptive and prescriptive model - the unit task Predictive Predicts the time it will take user to perform the tasks under analysis Descriptive Represents the way a user performs tasks on a system Prescriptive Guides the development of training programs and help systems GOMS: Definition GOMS models user’s behavior in terms of: Goals What the user wants to do. Operators Specific steps a user is able to take and assigned a specific execution time. Methods Well-learned sequences of subgoals and operators that can accomplish a goal. Selection Rules Guidelines for deciding between multiple methods. GOMS example GOAL: CLOSE-WINDOW . [select GOAL: USE-MENU-METHOD . MOVE-MOUSE-TO-FILE-MENU . PULL-DOWN-FILE-MENU . CLICK-OVER-CLOSE-OPTION GOAL: USE-CTRL-W-METHOD . PRESS-CONTROL-W-KEYS] For a particular user: Rule 1: Select USE-MENU-METHOD unless another rule applies Rule 2: If the application is GAME, select CTRL-W-METHOD Cognitive Complexity Theory - CCT - basic premises of goal decomposition provides more predictive power Two parallel descriptions: User - production rules of the form: if condition then action Device - generalized transition networks covered under dialogue models Example: editing with vi Production rules are in long-term memory - 4 rules in the text on page 425 User sees a mistake - Model contents of working memory as attribute-value mapping (GOAL perform unit task (TEXT task is insert space) (TEXT task is at 5 23) (CURSOR 8 7) Four rules to model inserting a space Notes on CCT Rules don’t fire in order written, may repeat Parallel model – rules can fire simultaneously Novice versus expert style rules Error behavior can be represented Measures Depth of goal structure Number of rules (more means interface more difficult to learn) Comparison with device description Problems with goal hierarchies description can be enormous a post hoc technique – risk is that it is defined by the computer dialog and not user expert versus novice Simple extensions possible goal closure (makes sure subgoal satisfied) eg. ATM example Linguistic notations User’s interaction with a computer is often viewed in terms of a language. Understanding the user's behaviour and cognitive difficulty based on analysis of language between user and system. Similar in emphasis to dialogue models Backus–Naur Form (BNF) Task–Action Grammar (TAG) Backus-Naur Form (BNF) Very common notation from computer science A purely syntactic view of the dialogue Terminals lowest level of user behaviour e.g. CLICK-MOUSE, MOVE-MOUSE Nonterminals ordering of terminals higher level of abstraction e.g. select-menu, position-mouse Example of BNF Basic syntax: nonterminal ::= expression An expression contains terminals and nonterminals combined in sequence (+) or as alternatives (|) draw line select line choose points choose one last point pos mouse ::= ::= ::= ::= ::= ::= select line + choose points + last point pos mouse + CLICK MOUSE choose one | choose one + choose points pos mouse + CLICK MOUSE pos mouse + DBL CLICK MOUSE NULL | MOVE MOUSE+ pos mouse Measurements with BNF Number of rules (not so good) Number of + and | operators Complications same syntax for different semantics no reflection of user's perception minimal consistency checking Task Action Grammar (TAG) Making consistency more explicit Encoding user's world knowledge Parameterised grammar rules Nonterminals are modified to include additional semantic features Consistency in TAG In BNF, three UNIX commands would be described as: copy ::= cp + filename + filename | cp + filenames + directory move ::= mv + filename + filename | mv + filenames + directory link ::= ln + filename + filename | ln + filenames + directory No BNF measure could distinguish between this and a less consistent grammar in which link ::= ln + filename + filename | ln + directory + filenames Consistency in TAG In TAG, this consistency of argument order can be made explicit using a parameter, or semantic feature for file operations. Consistency in TAG (cont'd) Feature Possible values Op = copy; move; link Rules file-op[Op] ::= command[Op] + filename + filename | command[Op] + filenames + directory command[Op = copy] ::= cp command[Op = move] ::= mv command[Op = link] ::= ln Other uses of TAG User’s existing knowledge Congruence between features and commands These are modelled as derived rules Physical and device models The Keystroke Level Model (KLM) Buxton's 3-state model Based on empirical knowledge of human motor system User's task: acquisition then execution. these only address execution Complementary with goal hierarchies Keystroke Level Model (KLM) lowest level of (original) GOMS six execution phase operators Physical motor: K - keystroking P - pointing H - homing D - drawing Mental M - mental preparation System R - response times are empirically determined. Texecute = TK + TP + TH + TD + TM + TR KLM example GOAL: ICONISE-WINDOW [select GOAL: USE-CLOSE-METHOD . MOVE-MOUSE-TO- FILE-MENU . PULL-DOWN-FILE-MENU . CLICK-OVER-CLOSE-OPTION GOAL: USE-CTRL-W-METHOD PRESS-CONTROL-W-KEY] compare alternatives: • • USE-CTRL-W-METHOD vs. USE-CLOSE-METHOD assume hand starts on mouse Architectural models All of cognitive models make assumptions about the architecture of the human mind. Problem Space Model Rational behavior is characterized as behavior that is intended to achieve a specific goal. This element of rationality is often used to distinguish between intelligent and machine-like behavior. In the field of artificial intelligence (AI), a system exhibiting rational behavior is referred to as a knowledge-level system. Problem Space Model A knowledge-level system contains an agent behaving in an environment. The agent has itself and its environment, including its own goals. knowledge It can perform certain and sense information about its changing environment. about actions The agent behaves in its environment, it changes the environment and its own knowledge. Problem Space Model We can view the overall behavior of the knowledge-level system as a sequence of environment and agent states as they progress in time. The goal of the agent is characterized as a preference over all possible sequences of agent/environment states. Problem Space Model Contrast this rational behavior with another general computational model for a machine, which is not rational. For example, it is common to describe a problem as the search through a set of possible states, from some initial state to a desired state. Problem Space Model The search proceeds by moving from one state to another possible state by means of operations or actions, the ultimate goal of which is to arrive at one of the desired states. Once a programmer has identified a problem and a means of arriving at the solution to the problem (the algorithm), the programmer then represents the problem and algorithm in a programming language, which can be executed on a machine to reach the desired state. Problem Space Model The architecture of the machine only allows the definition of the search or problem space and the actions that can occur to traverse that space. Termination is also assumed to happen once the desired state is reached. The machine does not have the ability to formulate the problem space and its solution, mainly because it has no idea of the goal. Problem Space Model It is the job of the programmer to understand the goal and so define the machine to achieve it. We can adapt the state-based computational model of a machine in order to realize the architecture of a knowledge-level system. The new the problem space model. computational model is Problem Space Model Thus, a problem space consists of states and a set of operations that can be performed on the states. a set of Behavior in a problem space is a two-step process. First, the current operator is chosen based on the current state and then it is applied to the current state to achieve the new state Summary Cognitive models attempt to represent users as they interact with the system. Most cognitive models do not deal with user observation and perception. Some techniques have been extended to handle system output, but problems persist. Summary Issues: Level of granularity Exploratory interaction versus planning