Cognitive Mechanisms of Learning

Anh Nguyen-Xuan

• Publisher: Wiley-ISTE
• Year: 2020
• Print ISBN:
  9781786305770
• Edition: 1
• eBook ISBN:
  9781119750475
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

Cognitive Mechanisms of Learning presents experimental research works on the issue of knowledge acquisition in Cognitive Psychology. These research works – initiated by groups of researchers with academic backgrounds in Philosophy, Psychology, Linguistics and Artificial Intelligence – explore learning mechanisms by viewing humans as information processing systems. Although the book is centered on research studies conducted in a laboratory, one chapter is dedicated to applied research studies, derived directly from the fundamental research works. Computer modeling of learning mechanisms is presented, based on the concept of ?cognitive architecture?. Three important issues – ?the methodology?, ?the achievements? and ?the evolution? – in the field of learning research are also examined.

1
Useful Concepts and Representation Formalisms

This chapter is split into two parts. In section 1.1, we define the main concepts used throughout this book, giving examples wherever possible. Section 1.2 contains a brief overview of the formalisms used to represent knowledge which an individual may possess at a given moment in their personal history. Some of these formalisms will be discussed in greater depth in later chapters as and when the need arises.

1.1. Useful concepts

In this section, we shall define the following concepts: information, information processing, problem, comprehension, short-term memory, long-term memory (LTM), episodic memory, working memory and knowledge. The latter is broken down into declarative and procedural knowledge.

In our approach, the human cognitive system is seen as an information processing system. It is therefore essential that we begin by clearly defining the concept of “information”.

1.1.1. Information

To define the concept of “information” and the companion concept of “redundancy”, let us start with an example: a conversation between two brothers, Andrew and Bradley, 15- and 17-year-old high school students.

• – Andrew: “Does the evening news program on ARTE usually start at 7, 7.30 or 8 pm?”
• – Bradley: “I don’t know exactly, but I know that it isn’t earlier than 7.30”.

In this case, Bradley’s answer is information, as it reduces Andrew’s uncertainty concerning the time of the program. One of the three initial possibilities has thus been eliminated.

Suppose that the conversation continues as follows:

• – Andrew: “So it isn’t at 7?”

Given the knowledge which Andrew possesses¹, his question is logically redundant, as it calls for a yes/no answer which will not allow him to determine whether the program is shown at 7.30 or 8 pm. It does not enable Andrew to reduce his uncertainty.

We offer the following definition of the concept of “information”.

DEFINITION.– Information. Take a situation located in space and time: a word, a scene or a signal (all instances of what we shall call “data”) is information if it provides one of the actors in the situation with knowledge, however insignificant, which the actor in question did not previously possess. Information is thus defined as a function of a situation, actors and the knowledge possessed by the latter.

1.1.2. Information processing

The expression “information processing” denotes a process that takes place over a period of time. During this time, a (structured) set of mechanisms is triggered to handle data coming from the external and/or internal environment. The internal environment considered here is long-term memory (LTM)², and the data obtained from this source is an individual’s declarative and procedural knowledge. The data being processed is held in the working memory (defined in section 1.1.5.4). The output of the process is a new piece of data, with the status of “information”. This information may or may not be consigned to LTM. It may also be held temporarily in the working memory as input for a subsequent process.

1.1.3. Problem

As in the case of “information” and “redundancy”, a “problem” is defined in relation to the knowledge of a given person in a given situation.

A “problem” is defined by a pair of elements: “situation/solver”.

A situation is considered “problematic” for a person if:

1. a) the person wishes to obtain a different situation, known as the “goal state”;
2. b) they cannot do so immediately;
3. c) they possess means which they can implement in order to achieve the goal state.

These characteristics imply that a given situation may or may not be problematic depending on the person’s knowledge.

Take the question “What’s four plus seven?”

• – For an adult, this question is not a problem as the solution is immediately available.
• – For a child of approximately 5 or 6 years, it is a problem: he or she will not find the solution immediately, but may attempt to find the solution using some form of counting process.
• – For a child of 2 years, this question is not a problem, but for a different reason: the child does not even understand the question.

This highlights two important points:

1. 1) A problem may be characterized by a “world state” in a given place at a given moment³. It may or may not constitute a “starting state” for a problem, depending on the “state of knowledge” of the person who identifies this world state as a problem situation, i.e. a situation from which he or she wishes to obtain a different state, i.e. the “goal state”.
2. 2) A situation will only constitute a problem for a person if that individual possesses a certain understanding, or comprehension, of the situation. This latter notion will be discussed in greater detail later.

Psychologists working on the subject of problem-solving distinguish between three different types of problems: “state transformation” problems, “rule induction” problems and “arrangement” problems.

A “state transformation” problem consists of defining a starting state and goal state using the same describers. Our “what’s four plus seven?” problem falls into this category. The starting and goal states use the same describers from the domain of arithmetic knowledge: numbers and arithmetic operators.

The aim of a “rule induction” problem is to identify relationships between a set of objects, as in cases of concept identification (see Chapter 4) or when completing a series: “1A, 2B, 3C, …?” (One possible answer is 4D).

A conception (or arrangement) problem is one in which the describers for the starting and goal states do not belong to the same domain. An anagram is an arrangement problem: the starting state is a word in a given language, the goal state is a word in the same language, but we do not know what word. However, we do know that the goal state must fulfill a certain number of criteria. Architecture problems (or “design” problems more generally) are arrangement problems⁴.

This distinction is primarily relevant for “simple” problems studied in a laboratory setting. Other problems combine multiple categories: the game of chess, for example, is both an arrangement and a transformation problem.

The definition of “problem” indicates that any situation in which a person wishes to attain a goal but cannot do so immediately is, for that person, a problem-solving situation. For example, a textual study, learning a poem by heart or planning an itinerary from one’s home to a new destination are all problem-solving situations.

Similarly, if a teacher has given a class, for example a history class, and wishes to know whether the students have understood the class, they may ask them to solve a problem: producing a summary of the class, responding to a questionnaire, etc.

Thus, the only way of judging the extent to which a learner has “learned” something is to put them in a “problem-solving” situation.

1.1.4. Comprehension

Taking th...

Table of contents

• Cover
• Table of Contents
• Foreword
• Acknowledgments
• Introduction
• 1 Useful Concepts and Representation Formalisms
• 2 Definition and Historical Overview
• 3 Learning to Solve a Problem
• 4 Learning a Concept from Examples of Concepts: Induction
• 5 Implicit Learning
• 6 The Role of Prior Knowledge in Constructing a Representation of a Problem
• 7 Acquiring Knowledge in a Specific Domain
• 8 Causal Learning
• 9 Symbolic Processing System Models in Cognitive Psychology
• Conclusion
• References
• Index
• End User License Agreement