Learning Theories PDF

Core Lecture 1.1 Learning Theories Sergey Sosnovsky TECHNOLOGIES FOR LEARNING Behaviorism  Behaviorism Learning = changes in either the form or frequency of observable performance/behavior Learning is accomplished when a proper response follows a stimulus ◦...

Core Lecture 1.1 Learning Theories Sergey Sosnovsky TECHNOLOGIES FOR LEARNING Behaviorism  Behaviorism Learning = changes in either the form or frequency of observable performance/behavior Learning is accomplished when a proper response follows a stimulus ◦ Stimulus: 4+2 =?; Response: 6 ◦ Learning = building this association, so that such a reaction is fast and reliable ◦ Learning happens through reinforcement of correct responses Many strategies are useful for building/strengthening stimulus-response associations: ◦ Instructional cues ◦ Practice to achieve fluency ◦ Reinforcement (mostly through positive feedback) Behaviorism No attempt to ◦ determine the structure of a student’s knowledge or ◦ assess involved mental processes or emotions Behaviorism is reliable and effective in facilitating learning that involves: ◦ discrimination (recalling facts), ◦ generalization (defining and illustrating concepts), ◦ association (applying explanations), ◦ chaining (automatically performing a specified procedure) Behaviorism does not work well for higher-order skills ◦ problem solving ◦ inference generation ◦ critical thinking…  Principles of behaviorist pedagogy Emphasis on producing observable and measurable outcomes in students ◦ behavioral objectives – what must a student be able to do ◦ task analysis into steps ◦ criterion-based assessment Pre-assessment of students to determine where instruction should begin Help mastering early steps before progressing to more complex levels of performance ◦ sequencing of instruction Use of reinforcement to impact performance ◦ tangible rewards ◦ informative feedback Use of cues/prompts and practice to ensure a strong stimulus-response association Linear programmed instruction and Skinner’s teaching machines Linear programmed instruction ◦ Carefully crafted sequence of small steps ◦ Learner receives automatic, immediate and regular reinforcement (positive feedback) ◦ Teaching happens through a response/reward mechanism ◦ The pace of learning is adjusted to suit the individual ◦ As a result: students are more interested, attentive, and learn efficiently by producing the desired behavior, "learning by doing"  Cognitivism A shift from the analysis of observable behaviour to the analysis of cognition Cognitive learning theories stress acquisition of knowledge and its internal structures Learning is a mental activity entailing internal coding and structuring by a learner ◦ new knowledge is connected to existing knowledge schemas in memory Learner is an active participant of the learning process Similarities with behaviourism: ◦ The goal of instruction is to efficiently communicate knowledge to students ◦ Knowledge is analyzed, decomposed, and simplified into basic building blocks ◦ Feedback is important as reinforcement (B) or support for connecting knowledge accurately (C) ◦ Learner analysis and task analysis are important  Principles of cognitivist pedagogy Emphasis on the active involvement of the learner in the learning process ◦ learner control over the learning process ◦ metacognitive training (e.g., self-planning, monitoring, and revising techniques) Subject knowledge and cognitive task analyses to identify and illustrate prerequisite-outcome relationships Emphasis on structuring and sequencing information to facilitate its processing ◦ use of cognitive strategies such as outlining, summaries, concept mapping, etc. Encourage students to make connections with previously learned material ◦ recall of prerequisite skills; use of relevant examples, analogies Constructivism Both, Behaviourism and Cognitivism are objectivistic: ◦ The world knowledge is real and external to the learner ◦ The goal of instruction is to map the world onto the learner Constructivism sees knowledge as an individual product of learning from experience Constructivism is a branch of cognitivism ◦ Learning is a mental activity of a learner gradually building up internal knowledge representation … but Constructivism differs from cognitivism ◦ Everyone’s knowledge is unique ◦ Mind is not a reference tool for the world, but the source of knowledge ◦ We create meaning, not acquire it  Constructivism Knowledge is fluid and changes as we learn ◦ Every learning effort is an interpretation of the current situation based on the history of interactions ◦ Learning best occurs in realistic settings when learning tasks are relevant to the students’ experience ◦ Memory is always under construction as a cumulative history of interactions The goal of instruction is not to ensure that learners know particular facts but rather that they can elaborate on and interpret information To be successful, meaningful, and lasting, learning must include three crucial factors: activity (practice), concept (knowledge), and culture (context) There is evidence that: ◦ Introductory learning is better supported by objectivistic approaches (behavioral and/or cognitive) ◦ Transition to the constructivist approaches should be made as learners acquire more knowledge and deal with complex and ill-structured problems Principles of constructivist pedagogy Emphasis on learner control and ability to manipulate information (actively using what is learned) Emphasis on the context in which the skills will be learned and subsequently applied (anchoring learning in meaningful contexts) Information should be presented in various ways (revisiting content at different times, in different contexts, for different purposes, and from different perspectives) Supporting problem-solving skills that allow learners to go “beyond the information given” (pattern recognition, finding an alternative solution, etc.) Assessment focused on transfer of knowledge and skills (presenting new problems and situations that differ from the conditions of the initial instruction)  Our cognitive architecture Interactions involve ◦ Perception and Attention ◦ Processing and Inference ◦ Memorization and Retrieval Long-term memory (LTM) ◦ Practically unlimited capacity and duration ◦ Knowledge is organized (schematic structures) Working memory ◦ Locust of conscious information processing ◦ Buffer between our sensors and LTM ◦ Severe limitations in capacity and duration  “Cognitive economy” principle Our brain organizes and acquires knowledge to minimize cognitive load ◦ External guidance < Existing LTM schema < Random search (trial and error) Better enhanced and frequently-activated LTM schemas are preferred ◦ Example: expert vs. novice thinking ◦ Sometimes these leads to erroneous outcomes ◦ Misconceptions during transfer ◦ Folk beliefs vs. Scientific knowledge For novices, external guidance can be the only source of executive cognitive function Optimal: combination of existing LTM schemas and external guidance Too little guidance is bad for novices ◦ They do not have enough LTM schemas and must engage the Trial and Error strategy Too much guidance is bad for experts ◦ Overlap between guidance and LTM can cause redundant mapping of external and internal structures and drain WM resources (expertise reversal effect) HCI Design of efficient learning environments should take into account main features and limitations of our cognitive architecture Ineffective design of instruction: ◦ Imposes redundant information processing ◦ Drains learner’s cognitive resources ◦ Hinders acquisition of knowledge Effective design of instruction: ◦ Promotes student-driven learning ◦ Triggers deep cognitive processes ◦ Result in active construction of new knowledge Cognitive Load Theory Our cognitive capacities are limited John Sweller (1980s) Some tasks are inherently difficult Design can be optimized to reduce bad cognitive load and free our resources for a good one Three sources of Cognitive Load Instructional design & interface Should be design, information modality decreased Inherent difficulty of a topic Can be 2+2 vs f''-f'-30f=0 managed Relation of a topic to existing Should be knowledge, activation of induced long-term memory (Learning) Sources of Extraneous CL related elements of information or representations that need to be processed simultaneously are separated in space and/or time too many new elements of information are introduced into WM and/or are introduced too fast to be successfully incorporated into LTM structures learners do not have necessary prior knowledge to deal with the situation, and/or instruction does not provide sufficient external guidance thus forcing learners to use random search procedures learners have sufficient prior knowledge that overlaps with provided external guidance thus requiring learners to mentally coordinate different representations of the same information  Multimedia Learning Theory Three main assumptions: Richard Mayer (1990s) ◦ There are separate channels for processing information of different modalities ◦ Channels have limited capacity (only a few chunks can be processed at a time per channel) ◦ Learning is an active process of filtering, selecting, organizing, and integrating information  Application: Principles for building Interactive Multimodal Learning Environments multimedia principle: understanding can be enhanced by engaging multiple channels (e.g., by the addition of non-verbal knowledge representations to verbal explanations) modality and verbal redundancy principle: when using multimedia, it is more effective to present verbal information in the auditory modality temporal contiguity and spatial contiguity principles: verbal and non-verbal information need to be synchronized in time and space coherence principle: students learn best when extraneous, distracting material is not included signaling principle: students learn best when they are shown exactly what to pay attention to on the screen Mastery Learning (Bloom, 1968)  Most students can learn everything that is being taught, if given sufficient time Principles Different students might require different amounts of time to achieve mastery Material should be broken into smaller objectives according to the prerequisite-outcome structure of the subject Students should be allowed to move at their own pace and take periodic formative assessment Students should not move to the next set of objectives before mastering the prerequisite set Students should receive often feedback on their current level of mastery Important contribution: Learning success depends not only on the student’s aptitude and efforts , but also on the quality of instruction (support of motivation and affect, well structured learning material, often and effective feedback) A typical mastery learning plan Relations to other methods of instruction Relations to behaviourism At its core mastery learning is a development of behaviourism and Skinner’s programmed instruction ◦ Learning in sequence = learning of most complex behavior rests upon learning a sequence of less complex component behaviors ◦ Focus on the role of corrective feedback ◦ Focus on smaller pieces of instruction as building blocks for implementing individualized learning ◦ Bulk of responsibility for the results of instruction lies with the teacher/instructor/environment and her/its effectiveness However, mastery learning ◦ Recognizes skills and objectives ◦ Stresses importance of motivation and affect  Mastery Learning in Cognitive Tutors / Task Sequencing Model exercises in terms of skills Sequence skills based on the prerequisite-outcome structure of the domain At each moment focus on a subset of skills Keep a student within this subset until mastery is reached Maintain individual models of skills Do not use skill-based feedback, instead use skillometer ◦ Use step-level feedback Use stealthed assessment = skills are assessed while a student solves exercises