The most highly requested skills for the workspace in the future are: oral communication (28 %), written communication (23%), collaboration (22%) and problem solving (19%).
The 21st-century skills are combination of cognitive (nonroutine problem solving, critical thinking, metacognition), interpersonal (social), and intrapersonal (emotional, self-regulation) skills. Skills such as creativity, originality, critical thinking, negotiation and complex problem-solving will increase in value.
The key to increasing productivity and business success is to use technology to help people work better, faster and smarter.
Acquiring knowledge requires comprehension as we try to understand and make sense of the world. Comprehension occurs in multiple ways such as exploring, listening and reading.
Richard Mayer proposed three cognitive processes that are required to make sense of explanatory material. The first requires selecting information from the passage. A second process is organizing the selected information into coherent body of knowledge. A final step in making sense of a text is integrating the knowledge derived from a text with previous knowledge stored in long-term memory. Mayer refers to these three processes of selecting, organizing and integrating as the SOI model of constructing knowledge.
Comprehension is much easier if we can integrate the material with knowledge stored in long-term memory. The best method to integrate ideas in the text with our own knowledge is to explain ideas in our own words.
The fourth aspect to comprehension that follows selecting, organizing and integrating knowledge is adding information. Adding information involves incorporating the new information into long-term memory rather than dismissing it as irrelevant or wrong.
Some say that the end goal of acquisition of knowledge is action. Larry Barselou argued that perception and action are central components of cognition. This action is cognitive theories is referred to as “embodies cognition”.
Object manipulation requires both an action and an object. We can talk about physical, virtual or mental manipulation of objects.
Categorization is important of all areas of knowledge. Categories consist of objects or events that we group together because we believe they are related.
Bruner, Goodnow and Austin listed five benefits of forming categories in their A Study of Thinking:
- Categorizing objects reduces the complexity of the environment
- Categorizing is the means by which objects of the world are identified.
- Categories reduce the need for constant learning.
- Categorizing allows us to select an appropriate action.
- Categorizing enables us to order and relate classes of objects and events.
Eleanor Rosch and Carolyn Mervis found that people generally agree on the typicality of category members. They hypothesized that typicality is determined by the extent to which category members shares attributes with other category members and does not share attributes with the members of other categories.
Gabriel Radvansky has conducted research that shows how events influence our memory. He defines “event boundaries” as transitions from one event to another.
Abstraction is double-edged sword. Three definitions of abstraction in the APA Dictionary of Psychology:
- An abstract entity exists only in the mind, separated from embodiment.
- Abstraction focuses on only some attributes of stimuli.
- An abstract idea applies to many particular instances of a category.
Concrete instances can be represented by sensory images. Images and verbal association are the two major forms of elaboration. The reason images are effective according to Paivio, is that image provides a second kind of memory code to support verbal memory. Paivio’s theory is called a dual coding theory.
Instances of categories are composed of attributes. Eleanor Gibson proposed that perceptual learning occurs through the discovery of features that distinguish one pattern from another.
A concrete way to represent a category is to remember many examples of category.
Sir Frederick Bartlett developed a general theory of organizing knowledge around “schema”. He defined schema as an active organization of past experience in which the mind abstracts a general cognitive structure to represent many particular instances of those experiences. A fundamental assumption of Bartlett’s theory is that all new information interacts with old information represented in the schema.
Roger Schank and Robert Abelson used the term script to refer to what we know about the sequence of events that make up routine activities.
How people learn how to comprehend, act, categorize and abstract. Robert Glushko, Paul Maglio, Teenie Matlock and Lawrence Barsalou refer to this type of acquisition as “cultural categorization”. Cultural categories exist for objects, events, mental states and other components of experience. Authors contract cultural categorization with institutional categorization. Institutions such as business, industry, law and science devoted considerable time and resources to develop classification system that serve their goals.
Systematic attempts to organize knowledge by using diagrams to create matrices, networks and hierarchies.
A matrix organizes categories along two dimensions. One dimension is “actions” and the other dimension is “objects”.
A person in a network does not have designated power over other members but may nonetheless have considerable influence on those members.
Networks consists of “nodes” joined by lines that are technically called “edges”. You can also use links instead of edges. The purpose of networks is to show how nodes are linked together. The density of connections is referred to as “clusters”. Two other key network concepts are the weight of a link and the length of a path.
A spreading activation model in which the length of each line (link) represents the degree of association between two concepts.
Frank Hillary and Jordan Grafman, showed that the brain imaging techniques demonstrated that information transfer times depend not only on the direct path between nodes but also on the availability of alternative, detour paths.
Niall Ferguson said that a hierarchy is just a special kind of network. The key to constructing a hierarchy is to start with the top node and always add nodes below it without connecting them laterally. There is only one path connecting any pair of nodes, which makes clear the chain of command and communication in organization.
The semantic network model illustrated how links can represent semantic relationship. Restriction can be placed on the links so the form a hierarchy as occurs in the hierarchical network model proposed by Alan Collins and Ross Quillian.
Ontologies are theories of knowledge based on hierarchical concepts. Michilene CHI constructed ontology to describe novice knowledge and its distinctions between entities, processes and mental states. Entities consists of objects and substances. Processes occur over time and can be sequential or emergent. Mental states are abstract, in one’s mind.
Sean Carroll describes how different perspectives influence how we view reality by referring to the air in a room. It can be a fluid or atoms.
The effective user of diagrams requires selecting the most appropriate one. This requires understanding how the diagrams differ from each other in their representation of information.
One distinguishes property is the diagram’s global structure. In a matrix, all the values of one variable share all the values of the other variable. In a network there is no predefined variable. In a hierarchy, the structure is organized into levels. Another distinguishes property concerns the pathways within a diagram. In the matrix there are no links so it doesn’t make sense to talk about moving along pathways. In the network there can be multiple paths between nodes. In the hierarchy there is only a single path from one node to another.
Language is a marvelous tool for communication, but is greatly overrated as a tool for thought.
We have seen examples of how hierarchies support reasoning but occasionally they can trick us. There is a strong tendency to distort judged locations to conform with the location of regions higher in the hierarchy. Another type of geographical bias is that people tend to judge geographical areas as more aligned than they actually are.
Organizing knowledge would be much easier if knowledge were perfect.
Words are responsible for a major proportion of ambiguous knowledge because they often have more than one meaning.
Challenges for Reasoning from Imperfect Knowledge:
- Ambiguous knowledge
- Conditional knowledge
- Contradictory knowledge
- Fragmented knowledge
- Inert knowledge
- Misclassified knowledge
- Uncertain knowledge
The spreading activation model proposed by Collins and Loftus assumes that activation spreads from a word to prime words with similar meaning.
Humans have a large knowledge base that allows them to avoid errors a computer might make when it follows rules that do not make sense.
Fragmented knowledge is not well connected to other knowledge stored in long-term memory. The knowledge-as-pieces theory contracts with those theories that propose knowledge involves a more coordinated use of concepts. Fragmentation occurs when learners create scientific explanation without concern for internal consistency and coherence.
High predictability makes life routine but is typically quite helpful.
Steven Sloman made a distinction between reasoning that is based on associations and reasoning that is based on rules.
We also have distinction between intuitive and analytical thinking. People apparently place too much faith in their intuitions and avoid cognitive effort as much as possible.
Heuristics are typically defined as strategies that often work but do not guarantee success.
Gerd Gigerenzer defines a “heuristic” in his book Risk Savy as a conscious strategy that ignores part of the information to make better judgments.
An alternative to nudging is boosting. Ralph Hertwig and Till Grune-Yanoff argue that nudging steers good decision whereas boosting empowers good decision. The goal of boosting is to create competencies through enhancing skills, knowledge and decision tools.
In his 1973 article “The Structure of Ill Structured Problems” Herbert Simon compared ill-structured with well-structured problems.
Most of the research on problem-solving continued to focus on well-structured problems.
The problem solver begins by constructing a representation to understand the problem by focusing on the goal, the constraints, and provided information.
Sandra Marshall in her Schemas in Problem solving is using definition that a schema is a memory organization that can: recognize similar experience; access a general framework that contains essential elements of those experiences; use the framework to draw interference, create goals, and develop plans; provide skills and procedures for solving problems in which the framework is relevant.
If the problem is unfamiliar and schematic knowledge is not activated, the problem solver must search for a solution by using general search for a solution by using general search methods such as means-end analysis and subgoals. Means-end analysis guides the search process by identifying moves that reduce the difference between the current problem state and the goal state. Subgoals are intermediate problem state between the initial and goal state.
Analogy is another general search process that works by attempting to find and then adapt a solution of a similar problem.
Another heuristic – working backwards – is more effective when there are many possibilities from the initial state and relatively few possibilities connected to the goal state.
Gestalt psychologists used the term “insight” to describe the sudden discovery of a correct arrangement of the parts following a succession of incorrect arrangements.
Experts tended to classify problems based on principles. Weaker problem solvers are sorting things based on story content.
Although expert-defined schemas are usually very helpful, they can occasionally constrain innovative solutions.
Erik Dane defines “cognitive entrenchment” as a high level of stability in knowledge schemas that can cause experts to be inflexible in their thinking.
A difference between design and nondesign tasks concerns the Gestalt concept of insight. Insight in Gestalt problems typically reveals the solution whereas insight in the more complex design problems typically reveals only an important step toward achieving solution.
Change over time, including transitions into new states.
Adrian Bejan. The key idea is that design in nature evolves in a manner that facilitates the movement of the substances that flow through it by forming tree-like structures. This principle applies to both physical systems such as rivers and biological systems such as the circulatory system.
Although nature is designed to facilitate flow, people are not. Conflicts often result in stalemates. Robin Vallacher, Peter Coleman, Andrzej Nowak and Lan Bui-Wrzosinska propose that the dynamical systems perspective provides a coherent theory of conflict resolution. A key aspect of their proposal is that conflict is influenced by attractors that are difficult to escape.
The theory proposed by Vallacher and his co-authors distinguishes among several types of attractors – positive attractors, negative attractors and latent attractors. There are two general ways in which an attractor can change and promote a search for strategies that better optimize problem solving. One scenario is to disassemble the attractor so it can be more productively reassembled. A second scenario for changing an attractor is to utilize a latent attractor that has receded to the background but can supplant the original attractor under the right set of conditions.
Implicit cognition is an idea, perception or concept that may be influential in the cognitive processes or behavior of an individual even though the person is not explicitly aware of it.
Dynamical systems are examples of complex systems composed of interacting components.
Pedro Dominges in his book The Master Algorithm is talking about how machine learning is all around us. The Analogizers in Domingo’s terminology develop methods to categorize patterns based on their similarity to other patterns.
The values of continuous measures can be plotted on continuous scale in which points near each other are more similar than points far from each other. These spatial measures of similarity have resulted in productive techniques for both machine learning and psychological modeling.
The Master Algorithm informs us that these spatial representations are an important tool for categorizing information in machine learning.
Classifiers emphasized a simple heuristic (take the best) that Gerd Gigerenzer describes in his book Risk Savy. This strategy bases decision on the best predicting features.
Bayes rule provides a method for revising beliefs based on new evidence. In states that the “Probability of a Hypothesis based on the Data” is proportional to the “Probability of obtaining the Data if the Hypothesis were correct” multiplied by the “Probability of the Hypothesis”.
Psychologists Charles Kemp and Joshua Tenebaum combined the Analogizer and Bayesian approaches by using Bayesian methods to evaluate different graph representation of similarity. The method is called hierarchical Bayesian analysis because it selects a structure for representing similarity and then arranges the data within the structure.
The Connectionists reverse-engineer what the brain does by adjusting the strength of connections between “neurons”. Connectionist learning differs from symbolic learning because concepts are distributed across “neurons” rather than represented by a one-to-one correspondence between concepts and symbols.
Knowledge for the Symbolists occurs by manipulating symbols that replace expressions with other expressions. Manipulating symbols to solve problems typically occurs by learning rules that combine different pieces of pre-existing knowledge.
There are several reasons for our perceptual advantage over machines. We learn to recognize objects in perceptually rich, dynamic, interactive environments whereas networks are trained on static images. A disadvantage for people, however, is capacity limits because we cannot visually perceive and represent all aspects of a scene.
Semanticscience Integrated Ontology (SIO). An important relation in molecular biology is “encodes”, defined by SIO as a relation between two objects in which the first object contains information that is used to produce the second object. There are two types of encoding used in genetics – “translated from” and “transcribed from”. Transcribed is a relation in which the information encoded in one object produces an exact or similar kind of object. Translated describes a relation in which a completely different kind of entity is generated.
We read by transferring words on a page into internal speech, which is often referred to as subvocalization. The text on the page (a textual entity) is translated into internal speech (a verbal language entity) and into meaning (propositions). The meaning may then be translated into visual images (an iconic mental representation).
Logic is the key ingredient of formal ontologies – a major tool of the information sciences.
And advantage of using SIO as a biomedical foundation for cognition is that its four top-level categories – objects, process, attribute and relation – correspond a fundamental component of knowledge in psychology.
A physical entity has a location in space/time and is partitioned into “object” and “process”. An “object” is a physical thing like a chair or glass of water. A “process” is an action that occurs over time like a lecture. An abstract entity cannot exist at particular place in space/time without some physical encoding or embodiment. Subcategories of “abstract” include “quantity”, “attribute”, “relation” and “proposition”.
A “psychological process” is one type of “process” that includes “perception”. “Seeing” and “hearing” are subclasses of “perception”. “Looking” is an intentional act of “seeing” and “listening” is an intentional act of “hearing”.
Formal ontologies use hierarchical network as property of hierarchies to make deductions. Logic can also be used to represent knowledge and as such to identify contradictions.
George Miller proposed that the capacity of short-term memory varies from five to nine chunks of information.
We can partion semantic knowledge into taxonomic and thematic relations. Taxonomic relations capture similarities based on shared features. Thematic categories capture co-occurences in events or scenarios. Taxonomic relations consist of relatively static features. In contrast, thematic relations consist of action features. The co-occurrence of objects in thematic relations provides context in which objects occur.
The current trend in AI is to combine more computational power with more training data – doing more with more. In contrast, people do more with less.
Alison Gopnik is collaborating with her AI colleagues to build a system names MESS for model-building, exploratory, social learning system.
There is an attempt to develop a standard model of the mind to provide a common framework for unifying model of mind to provide common framework for unifying AI, cognitive psychology, cognitive neuroscience and robotics.
A standard model would provide a shared foundation for comparing human and machine reasoning. Robotics is one of the four domains included in standard model. One of the most fundamental requirements for both robots and people is that ability to navigate in a spatial environment.
A standard model of the mind includes Declarative Long-term Memory and Procedural Long-term Memory that are two-way connected with Working Memory, that is filled with Perception and activate Motor.
Understanding requires learning words and concepts, interpreting indirect speech, managing ambiguity and incongruity and pursuing implications.
Physical Turing test could integrate four major aspects of AI research: perception, action, language and reasoning.
What is the distinction between component-dominant and interaction-dominant systems? The relationships between the components in component-dominant system are primarily sequential. Language is an example of sequential process because one world follows another word.
The core ideas of the Next Generation Science Standards are:
- Have a broad importance across multiple sciences or engineering disciplines or be a key organizing principle of a single discipline.
- Provide a key tool for understanding or investigating more complex ideas and solving problems.
- Relate to the interests and life experiences.
- Be teachable and learnable over multiple grades at increasing levels of depth and sophistication.
The science standards are goals that reflect what a student should know; they do not dictate the manner or methods by which the standards are taught. The framework identifies crosscutting themes. The themes include patterns, establishing cause and effect, identifying structure and function, and studying the stability and change of systems.
Michael Jacobson and Manu Kapur are leaders in the study of instruction on complex systems.
Climate change is only one topic that would benefit from more instruction on complex systems.
- Structures refer to the physical features of the system such as the number and name of the variables and how the variables are connected to each other.
- Processes refer to the dynamics of the system such as self-organization, the causal nature of relationships, and emergence.
- States refer to how complex systems exists in the world as a result of shifts to existing structures and processes.
Computational and Mathematical Thinking
Computational thinking is a way of solving problems, designing systems and understanding human behavior that draws on concepts fundamental to computer science. Computer science is the study of computers and algorithmic processes including their principles, hardware, software and applications to society. Coding is the practice of developing a set of directions that a computer can understand and execute.
Characteristic of Computational Thinking
Learning how to program a computer improves a variety of cognitive skills. The instruction improved reasoning, creative thinking, metacognition (managing cognition), spatial skills, and mathematical skills.
In the Pursuit of the Unknown: 17 Equations that changed the world by Ian Stewart:
- Phythagoras’s Theorem
- Newton’s Law of Gravity
- Square Root of Minus One
- Formula for Polyhedra
- Normal Distribution
- Wave Equation
- Fourier Transform
- Navier-Stokes Equation
- Maxwell’s Equation
- Law of Thermodynamics
- Schrodinger’s Equation
- Information Theory
- Chaos Theory
- Black-Scholes Equation
Mathematical and computational thinking both includes: problem solving, modeling, analyzing and interpreting data and statistic and probability.
“The man who graduates today and stops learning tomorrow is uneducated the day after.” – Newton D. Baker
The current lack of confidence in institutions and in the acceptance of the value of lifelong learning provides a clear opportunity for leaders in higher education.
Leonardo Da Vinci demonstrated in the 15th Century cognitive skills that will endure forever in their importance. aspiration to improve in ourselves, curiosity and intense observation.
 In the book on page 233