What is Cognitive Computing?

Cognitive computing is a branch of computing that aims to mimic human thought processes
using artificial intelligence (AI), machine learning (ML), natural language processing (NLP),
and data analysis techniques. Unlike traditional computing systems that are programmed to
follow strict rules, cognitive systems are designed to learn, reason, and interact naturally
with humans, enabling them to make decisions or provide insights in complex and uncertain
situations.

It is widely used in fields like healthcare, finance, education, cybersecurity, and customer
service, where systems need to analyze large volumes of unstructured data and assist in
decision-making.

Key Elements of Cognitive Computing

1. Machine Learning (ML)

o Meaning: Enables systems to learn from data patterns and improve

performance over time without explicit programming.
o Role in Cognitive Computing:
 Helps systems adapt to new data and scenarios.
 Example: A medical cognitive system learning to identify diseases
from past diagnostic images.

2. Natural Language Processing (NLP)

o Meaning: Technology that allows computers to understand, interpret, and

respond to human language in spoken or written form.
o Role:
 Helps in understanding unstructured data like emails, social media, or
research papers.
 Example: Chatbots, voice assistants (like Siri, Alexa), and automated
customer support.

3. Reasoning and Problem-Solving

o Meaning: Ability to use logic and evidence to reach conclusions and

recommendations.
o Role:
 Supports decision-making in uncertain or incomplete data situations.
  Example: In finance, reasoning systems can suggest investment
strategies based on trends and risks.

4. Adaptive Learning

o Meaning: Systems continuously update themselves as new information

becomes available.
o Role:
 Keeps the system relevant in dynamic environments.
 Example: Fraud detection systems adapt to new fraud patterns.

5. Human–Computer Interaction (HCI)

o Meaning: Interfaces that allow users to interact with systems in a natural

and intuitive way (text, speech, gesture).
o Role:
 Builds trust and usability by making machines communicate like
humans.
 Example: Virtual assistants that understand voice commands.

6. Contextual Awareness

o Meaning: Ability to understand the context of data—time, location,

environment, and intent.
o Role:
 Improves accuracy of recommendations.
 Example: A cognitive travel assistant suggesting routes based on
traffic, weather, and user preferences.

7. Big Data Integration

o Meaning: Cognitive systems process both structured (databases) and

unstructured (text, images, videos) data.
o Role:
 Extracts useful patterns and insights from massive datasets.
 Example: Healthcare systems analyzing patient records, genetic data,
and medical journals to recommend treatments.
 8. Automation and Robotics

o Meaning: Using AI-powered decision-making to automate tasks that typically

require human intelligence.
o Role:
 Enhances efficiency and reduces human error.
 Example: Automated financial advisors (robo-advisors) in banking.

Conclusion

Cognitive computing is not just about processing information—it’s about thinking,

reasoning, and interacting like humans. Its core elements—machine learning, NLP,
reasoning, adaptive learning, contextual awareness, and human-computer interaction—
work together to build intelligent systems that learn, adapt, and assist in decision-making.
What is Cognition?
Cognition refers to the mental processes involved in acquiring, processing, storing, and
using knowledge.
It’s essentially how the brain enables us to think, learn, remember, problem-solve, and
make decisions.

In simple terms: Cognition is “the way we know and understand the world.”

Key Elements of Cognition

1. Perception

 How we receive information through senses (sight, hearing, touch, etc.).

 Example: Recognizing a friend’s voice in a crowded room.

2. Attention

 The ability to focus on specific information while ignoring distractions.

 Example: Reading a book in a noisy café.

3. Memory

 Storing and retrieving information.

o Short-term/working memory → holds info temporarily (like a phone
number).
o Long-term memory → stores knowledge, experiences, skills.

4. Learning

 Process of acquiring new information or modifying existing knowledge.

 Example: Learning to ride a bicycle or understand a new programming language.
5. Language

 Ability to understand, produce, and use words for communication.

 Example: Writing an essay or speaking in conversation.

6. Reasoning & Problem-Solving

 Using logic, rules, and strategies to solve problems or make decisions.

 Example: Planning a trip within a budget.

7. Decision-Making

 Choosing between alternatives based on reasoning, emotions, and goals.

 Example: Selecting a career path after graduation.

8. Metacognition

 “Thinking about thinking” — self-awareness of one’s own cognitive processes.

 Example: Realizing you need to change your study method because you’re not
remembering well.

Approaches to Studying Cognition

1. Cognitive Psychology → Focuses on mental processes using experiments.
2. Neuroscience → Studies brain structures (neurons, networks) behind cognition.
3. Artificial Intelligence (Cognitive Computing) → Tries to replicate human cognition in
machines.
4. Educational Psychology → Applies cognitive principles to improve learning and
teaching.

Importance of Understanding Cognition

 Helps in education (better teaching/learning methods).
 Useful in AI & cognitive computing (building intelligent systems).
 Improves mental health treatments (understanding disorders like memory loss,
ADHD, dementia).
 Supports decision-making and problem-solving in daily life.
Concept of Complex System
A complex system is a system made up of many interconnected parts (called components,
agents, or subsystems) that interact with each other in non-linear ways.

 The behavior of the whole system cannot be understood just by studying its parts
individually.
 It often shows emergent properties (new behaviors or patterns that arise from the
interactions of parts).
 Complex systems are dynamic, adaptive, and unpredictable.

Key characteristics of a complex system:

1. Many interacting components.

2. Non-linear relationships (small changes can lead to big effects).
3. Emergent behavior (patterns or behaviors not present in individual parts).
4. Adaptability (system can evolve and learn).
5. Self-organization (system can arrange itself without central control).

Examples of Complex Systems (with

Explanation)

1. The Human Brain

 Why complex?
o It consists of billions of neurons (components) interconnected through
trillions of synapses.
o Each neuron follows simple electrical rules, but together they create
intelligence, emotions, creativity, and consciousness.
 Emergent property:
o Consciousness and thought processes are emergent properties of neural
interactions.
 Example:
o When learning a new language, multiple brain regions interact (memory,
speech, hearing) and reorganize connections over time.

2. Ecosystem (e.g., Forest or Ocean System)

  Why complex?
o Ecosystems have plants, animals, microorganisms, soil, water, and climate
interacting.
o Species depend on each other (food chain, symbiosis).
 Emergent property:
o Biodiversity, food webs, and resilience to natural changes.
 Example:
o If a predator species (like wolves) is removed from an ecosystem, the prey
population may grow uncontrollably, leading to overgrazing and collapse of
vegetation — a small change causing large-scale effects.

3. Financial Market

 Why complex?
o Millions of investors, banks, and companies interact, influenced by global
events, policies, and emotions.
o Decisions are not purely rational; they are affected by psychology and
external factors.
 Emergent property:
o Market crashes, bubbles, and unpredictable price fluctuations.
 Example:
o In 2008, a small trigger in the US housing market caused a global financial
crisis, showing how interconnected and fragile the system is.

4. Internet / World Wide Web

 Why complex?
o Billions of devices (computers, servers, phones) connected worldwide.
o Data flows dynamically between networks, websites, and users.
 Emergent property:
o Social networks, viral trends, cyber-attacks, and collective knowledge (like
Wikipedia).
 Example:
o A single tweet or post can trigger global social movements (e.g., #MeToo),
showing the unpredictable collective behavior of users.
Artificial Intelligence to Cognitive Computing
Transforming Artificial Intelligence (AI) into Cognitive Computing involves enhancing AI systems so
they can mimic human thought processes, such as reasoning, learning, perception, and decision-
making, rather than just performing rule-based tasks. Cognitive computing systems aim to
understand context, interpret unstructured data, and adapt over time, making them more human-
like in intelligence compared to traditional AI.

1. From AI to Cognitive Computing

 AI (Artificial Intelligence): AI systems are designed to perform specific tasks like

image recognition, speech recognition, or playing games. They rely on algorithms,
rules, and pattern recognition.
 Cognitive Computing: Cognitive systems go beyond simple task execution—they
simulate human reasoning by understanding natural language, learning from
experience, and providing insights in uncertain situations.

Key Steps in Transformation:

1. Integration of Natural Language Processing (NLP)

o AI systems need to process and understand human language to interact
meaningfully.
2. Incorporation of Machine Learning (ML) and Deep Learning
o Cognitive systems learn from new data and improve over time rather than
being static.
3. Contextual Understanding and Reasoning
o Cognitive systems analyze context, draw inferences, and make decisions like
humans do.
4. Handling Unstructured Data
o Unlike AI, which often relies on structured data, cognitive systems can
analyze text, images, videos, and other unstructured sources.

2. Examples

Example 1: IBM Watson

 AI Function: Basic AI can answer questions using rule-based algorithms.

 Cognitive Computing Transformation: Watson can read, understand, and analyze
large volumes of unstructured data (medical journals, reports). It provides evidence-
based treatment recommendations for doctors, learns from outcomes, and adapts its
reasoning over time.

Example 2: Personal Assistants (e.g., Siri, Google Assistant evolving into cognitive
systems)
  AI Function: Simple voice commands and keyword-based actions.
 Cognitive Computing Transformation: Advanced assistants understand context,
sentiment, and user preferences. For instance, they can suggest reminders based on
your schedule, habits, or even mood, and they learn from past interactions to improve
responses.

Summary:
Transforming AI into cognitive computing is about adding human-like understanding,
reasoning, and learning capabilities to AI systems. While AI performs tasks, cognitive
computing interprets, learns, and reasons, enabling smarter, context-aware decisions.
Uses of Cognitive Computing System
1. Healthcare and Medical Diagnosis

 Use: Assisting doctors in diagnosing diseases and recommending treatments.

 Explanation: Cognitive systems like IBM Watson Health can analyze vast amounts
of medical data, research papers, patient history, and clinical trials. They help doctors
identify patterns, recommend evidence-based treatments, and predict patient
outcomes.
 Example: Suggesting personalized cancer treatment plans based on the patient’s
medical records and latest research.

2. Financial Services

 Use: Fraud detection, risk assessment, investment advisory, and customer support.
 Explanation: Cognitive systems can detect unusual patterns in transactions to flag
fraud, assess risks for loans or investments, and provide personalized financial advice.
They process both structured and unstructured financial data in real-time.
 Example: Banks using cognitive systems to monitor transactions for suspicious
activity or to provide AI-driven wealth management advice.

3. Customer Service and Support

 Use: Intelligent chatbots and virtual assistants that improve customer interaction.
 Explanation: Cognitive systems understand natural language, sentiment, and context,
allowing them to handle queries, complaints, and requests more effectively than
traditional AI. They can escalate complex issues to human agents when needed.
 Example: Chatbots on e-commerce websites that resolve issues, provide
recommendations, or guide users through complex processes.

4. Retail and Marketing

 Use: Personalized recommendations, inventory management, and demand forecasting.

 Explanation: Cognitive computing systems analyze customer behavior, purchase
history, and social media interactions to offer personalized product recommendations,
optimize inventory, and predict future demand.
 Example: Online stores like Amazon or Netflix suggesting products or movies based
on user preferences and viewing patterns.
5. Education and Personalized Learning

 Use: Adaptive learning platforms and student performance analysis.

 Explanation: Cognitive systems can assess students’ learning styles, strengths, and
weaknesses, and provide customized learning paths. They also help teachers by
analyzing performance data and predicting students who may need extra support.
 Example: Platforms recommending specific lessons or exercises to improve
understanding of complex topics for each student.

6. Cybersecurity

 Use: Threat detection, intrusion prevention, and risk analysis.

 Explanation: Cognitive systems analyze network traffic, logs, and historical attack
data to detect anomalies and predict cyber threats. They adapt to new attack patterns
faster than traditional security systems.
 Example: AI-driven cybersecurity tools identifying suspicious activity before it leads
to a breach.

7. Human Resources and Recruitment

 Use: Candidate screening, performance evaluation, and employee engagement.

 Explanation: Cognitive systems can analyze resumes, social profiles, and past
performance data to shortlist candidates. They also help monitor employee
satisfaction and predict retention risks.
 Example: Tools that suggest best-fit candidates for a role by analyzing both skills and
cultural fit.

8. Supply Chain and Logistics

 Use: Optimizing operations, demand forecasting, and predictive maintenance.

 Explanation: Cognitive systems analyze production data, supplier performance, and
market trends to optimize supply chains, reduce costs, and prevent disruptions.
 Example: Predicting delays in shipments and suggesting alternative routes or
suppliers.

9. Legal and Compliance

 Use: Contract analysis, legal research, and compliance monitoring.

 Explanation: Cognitive systems can quickly review large volumes of legal
documents, extract key clauses, and identify compliance risks. They reduce manual
effort and improve decision-making in legal operations.
 Example: An AI tool analyzing contracts to flag non-compliance with regulations.
10. Smart Cities and IoT

 Use: Traffic management, energy optimization, and public safety.

 Explanation: Cognitive systems process data from sensors, cameras, and IoT devices
to make cities smarter and more efficient. They can optimize traffic flow, reduce
energy consumption, and predict emergencies.
 Example: Traffic systems that adjust signal timings in real-time based on traffic
density.
Gaining Insights from Data in Cognitive Computing
What It Means

In cognitive computing, data is transformed into actionable knowledge. Unlike traditional

systems that rely on pre-defined rules or structured databases, cognitive systems can:

 Understand unstructured data (text, images, videos, social media posts, emails).
 Analyze large volumes of diverse data quickly.
 Identify patterns, trends, and correlations that humans might miss.
 Provide recommendations, predictions, or explanations based on insights.

Essentially, cognitive computing turns raw data into meaningful, context-aware insights for
decision-making.

2. How Cognitive Computing Gains Insights from Data

Step 1: Data Collection

 Collect data from multiple sources:

o Structured: Databases, spreadsheets, sensors
o Unstructured: Text documents, emails, social media, images, videos

Step 2: Data Understanding

 Cognitive systems use Natural Language Processing (NLP), image recognition, and
speech recognition to understand content and context.
 Example: Analyzing customer complaints to determine common pain points.

Step 3: Pattern Recognition and Analysis

 Machine learning algorithms identify patterns, trends, and relationships within the
data.
 Example: A healthcare system finding correlations between patient symptoms and
rare diseases.

Step 4: Reasoning and Interpretation

 Cognitive systems interpret the patterns and link them to business, scientific, or
operational objectives.
 Example: Predicting equipment failure in a factory based on sensor data trends.

Step 5: Insight Generation

 The system summarizes findings and generates actionable recommendations.

 Example: Suggesting personalized treatment plans in healthcare or marketing
strategies in retail.
Step 6: Continuous Learning

 Insights improve over time as the system learns from new data and outcomes,
making predictions more accurate.

3. Examples of Gaining Insights from Data

Example 1: Healthcare

 Cognitive system analyzes medical records, lab reports, and research papers.
 Identifies patterns in symptoms, treatment effectiveness, and genetic data.
 Provides doctors with evidence-based recommendations for patient treatment.

Example 2: Retail

 Analyzes customer purchase histories, social media trends, and seasonal patterns.
 Gains insights into customer preferences, product demand, and marketing
effectiveness.
 Helps companies make data-driven decisions for promotions and inventory
management.

Example 3: Financial Services

 Cognitive system processes transaction data, market reports, and news articles.
 Detects unusual patterns that indicate fraud or market risk.
 Suggests investment strategies and alerts analysts to potential threats.

4. Benefits of Gaining Insights from Data in Cognitive Computing

1. Improved Decision-Making: Supports decisions with evidence and context.

2. Predictive Capabilities: Anticipates future trends, failures, or opportunities.
3. Personalization: Offers tailored recommendations to users or customers.
4. Efficiency: Reduces the time needed to analyze large, complex datasets.
5. Adaptability: Continuously learns from new data to improve insights.
Bringing Data into a Cognitive System
Cognitive computing systems rely heavily on data because they learn, reason, and make
decisions based on the information they receive. Bringing data into the system is a critical
first step, and it involves multiple layers: collection, integration, pre-processing, and
storage.

1. Sources of Data

Cognitive systems use both structured and unstructured data from various sources:

A. Structured Data

 Organized data in rows and columns.

 Examples: Databases, spreadsheets, ERP systems.
 Uses: Easy to process using traditional queries and machine learning algorithms.
 Example: Customer purchase history stored in a CRM database.

B. Unstructured Data

 Data without a predefined format.

 Examples: Emails, social media posts, images, audio, video, PDFs, scientific articles.
 Uses: Requires advanced processing (NLP, computer vision) to extract meaning.
 Example: Analyzing customer sentiment from social media comments.

C. Semi-Structured Data

 Partially organized, e.g., JSON, XML files.

 Example: Sensor logs from IoT devices or online forms.

2. Methods of Bringing Data into a Cognitive System

A. Data Integration

 Combines data from multiple sources into a unified view.

 Cognitive systems can handle heterogeneous data, connecting databases, web
sources, sensors, and social media feeds.
 Example: A healthcare system integrates patient records, lab results, and research
papers to get a complete patient profile.

B. Data Pre-processing

 Raw data often contains errors, duplicates, or irrelevant information.

 Pre-processing ensures data quality and consistency:
 o Cleaning: Removing errors and duplicates
o Normalization: Standardizing formats
o Transformation: Converting data to a usable format for analysis
 Example: Converting all dates in a dataset to a single format for consistency.

C. Data Storage

 Cognitive systems require scalable storage solutions to handle large volumes of data:
o Cloud storage
o Data lakes
o Big Data platforms
 Example: Retailers store customer interactions, purchase histories, and social media
data in a central data lake for analysis.

D. Real-Time Data Streaming

 Cognitive systems can process live data to provide real-time insights.

 Sources include IoT sensors, stock market feeds, or social media streams.
 Example: Smart city traffic sensors provide live updates to optimize traffic flow.

3. Importance of Bringing Data Properly

 Accuracy: Good data ensures reliable insights.

 Context: Cognitive systems need context to reason effectively.
 Scalability: The system must handle growing volumes of data without performance
loss.
 Learning: High-quality and diverse data enables better machine learning and
continuous improvement.

4. Real-World Example

Healthcare Cognitive System

1. Data Sources: Patient medical history, lab results, imaging data, research papers.
2. Integration: Combine all patient-related data into a single system.
3. Pre-processing: Clean errors in lab results, standardize medical terminologies,
anonymize patient data for privacy.
4. Storage: Use cloud-based platforms for large-scale storage.
5. Outcome: Cognitive system analyzes all the data to recommend personalized
treatment options and predict outcomes.

5. Summary
 Bringing data into a cognitive system is the foundation of all cognitive computing
applications.
 It involves:
1. Collecting data from structured, unstructured, and semi-structured sources.
2. Integrating data from multiple sources.
3. Pre-processing data to ensure quality.
4. Storing and managing data efficiently.
 Proper data handling ensures that cognitive systems can learn, reason, and generate
actionable insights effectively.
Language role in the cognitive process
Language plays a pivotal role in the cognitive process because it is the primary medium
through which humans think, communicate, and structure knowledge. Cognitive processes
include perception, memory, reasoning, problem-solving, and learning. Language does not
merely express thought—it actively shapes and organizes it. Here’s a detailed explanation:

1. Language as a Tool for Thought

Language allows humans to categorize, conceptualize, and manipulate abstract ideas. The
words and syntax we use influence how we perceive and process information. Cognitive
psychologists often argue that thought is closely tied to linguistic structures—a concept
known as the linguistic relativity hypothesis.

Example 1: Problem-Solving and Reasoning

 When solving a mathematical problem, internal verbalization (“I need to add these
numbers first, then subtract…”) helps structure the sequence of steps.
 People who talk themselves through a task often perform better because language
structures working memory, allowing them to plan, analyze, and evaluate.

2. Language as a Medium for Memory and Learning

Language enables humans to store and retrieve information efficiently. Words act as
mental labels that categorize experiences, making it easier to recall and communicate
complex ideas.

Example 2: Learning a New Concept

 A child learning the concept of “gravity” understands it better when the teacher uses
precise language: “Objects fall towards the Earth due to gravitational force.”
 The verbal explanation helps the child link abstract ideas to concrete words,
enhancing comprehension and retention. Without language, understanding such
concepts would be much slower or fragmented.
Defining Objectives, Defining Domain, Understanding the
users and defining their attributes, defining questions and
exploring insights, training and testing of cognitive
computing systems.
1. Defining Objectives

This is the first and most crucial step in cognitive computing projects. Here, you clearly
specify what you want the system to achieve. Objectives should be:

 Specific: Define precisely what the system is expected to do (e.g., predict customer
churn, detect fraud).
 Measurable: Ensure outcomes can be quantified (e.g., accuracy, response time).
 Achievable: Make sure goals are realistic given available resources and technology.
 Relevant: Align with organizational needs or user requirements.
 Time-bound: Set timelines for achieving results.

Example: “Develop a cognitive system that predicts equipment failure in manufacturing with
90% accuracy within 6 months.”

2. Defining Domain

The domain refers to the specific area of knowledge or application the system will operate in.
Cognitive computing systems rely on domain knowledge for reasoning and decision-making.

 Domain Scope: Identify boundaries and limitations of the domain (e.g., healthcare,
finance, retail).
 Data Requirements: Determine the type of data available in the domain.
 Knowledge Representation: Decide how domain knowledge will be stored
(ontologies, rules, knowledge graphs).

Example: In a healthcare domain, the system may focus on diagnosing diseases based on
patient symptoms and medical history.

3. Understanding Users and Defining Their Attributes

Cognitive systems must understand and adapt to users. This involves identifying who will
interact with the system and their characteristics:

 User Profiles: Age, profession, expertise level, preferences.

 Behavior Patterns: How users interact with technology or data.
 Cognitive Attributes: Decision-making style, biases, emotional states.
  Needs & Goals: What users expect from the system and what problems they want to
solve.

Example: A personalized learning system may adapt content based on a student’s prior
knowledge, learning pace, and engagement levels.

4. Defining Questions and Exploring Insights

Here, the focus is on what the system should answer or help discover:

 Problem Formulation: Convert business or research problems into cognitive

questions.
 Data Exploration: Analyze available data to discover patterns, relationships, and
insights.
 Hypothesis Testing: Use data to validate assumptions.
 Insight Generation: Transform raw data into actionable intelligence for decision-
making.

Example: In retail, questions like “Which products are likely to trend next month?” guide the
system to find insights from sales and social media data.

5. Training and Testing of Cognitive Computing Systems

Cognitive systems often use machine learning or AI models, which require training and
testing to ensure accuracy and reliability.

 Data Preparation: Clean, normalize, and structure data.

 Training Phase: Feed historical data to the system so it can learn patterns and
relationships.
 Testing/Validation: Evaluate performance using unseen data to avoid overfitting.
 Feedback Loop: Continuously refine the system based on performance and new data.

Example: A fraud detection system is trained with historical transaction data and tested on
recent transactions to validate its ability to detect fraudulent activity accurately.

Summary Flow

1. Define objectives → What the system should achieve.

2. Define domain → Where the system operates.
3. Understand users & attributes → Who interacts with the system.
4. Define questions & explore insights → What the system should answer.
5. Train and test → Make the system intelligent, adaptive, and reliable.
Cognitive models
1. Goal and Task Hierarchies

Definition:

Goal and task hierarchy models represent human behavior in terms of goals, subgoals, and
the tasks required to achieve them. They are primarily used to understand problem-solving
and decision-making processes.

Key Points:

 Goals: High-level objectives that a person wants to achieve.

 Subgoals: Smaller, manageable objectives that contribute to the main goal.
 Tasks/Actions: Concrete steps performed to achieve subgoals.
 Can model sequential or parallel actions depending on task requirements.

Example:

 Goal: Plan a trip.

 Subgoals:
1. Book transportation
2. Reserve hotel
3. Plan itinerary
 Tasks: Search flights online, compare hotel prices, create a daily schedule.

Applications:

 User interface design: Predicting user actions and improving usability.

 Workflow automation: Breaking complex goals into tasks for AI planning systems.

2. Linguistic Models

Definition:

Linguistic cognitive models focus on how humans understand, produce, and process
language. They simulate language comprehension, generation, and reasoning in cognitive
tasks.

Key Points:

 Represent syntax, semantics, and pragmatics of language.

 Often based on grammars, semantic networks, or discourse structures.
 Can explain how language influences thought, reasoning, and decision-making.

Example:
  Understanding a sentence like: “If it rains, the match will be canceled.”
 Linguistic models represent the conditional structure and derive logical consequences.

Applications:

 Natural Language Processing (NLP) systems.

 Cognitive AI chatbots and virtual assistants.
 Language learning tools and speech recognition systems.

3. Physical and Device Models

Definition:

Physical and device cognitive models represent how humans interact with the physical
world or devices. These models focus on motor skills, perceptual-motor coordination,
and interface interactions.

Key Points:

 Incorporate human sensorimotor capabilities, such as hand-eye coordination.

 Can predict reaction times, error rates, and physical constraints during interaction.
 Useful for designing ergonomic and user-friendly systems.

Example:

 Modeling how a pilot uses cockpit controls:

o How long it takes to reach a switch.
o How different layouts affect accuracy and efficiency.

Applications:

 Human-Computer Interaction (HCI) and user interface design.

 Robotics: Modeling human-like motion in robot control.
 Simulation training for pilots, surgeons, or operators.

4. Architectural Models

Definition:

Architectural cognitive models describe the internal structure of the mind, including
memory systems, attention, and information processing pathways. They often integrate
multiple cognitive functions into a coherent framework.

Key Points:
  Define components, their functions, and interactions.
 Can include working memory, long-term memory, attention systems, and
executive control.
 Serve as blueprints for simulating human cognition in computational systems.

Example:

 ACT-R (Adaptive Control of Thought – Rational):

o Integrates declarative memory, procedural knowledge, and perception-action
modules.
o Used to simulate learning, problem-solving, and human-computer interaction.

Applications:

 Cognitive architectures in AI and robotics.

 Understanding human cognitive limits and designing cognitive support systems.
 Modeling attention, learning, and reasoning processes in complex tasks.

Summary Table

Model Type Focus Components / Example Application

Features
Goal & Task Human goals Goals, subgoals, Planning a trip Workflow
Hierarchies & subgoals tasks automation, UI
design
Linguistic Language Syntax, semantics, Understanding NLP, chatbots,
processing discourse conditionals language learning
Physical & Human-device Motor skills, Pilot using HCI, ergonomics,
Device interaction perceptual-motor cockpit controls robotics
coordination
Architectural Internal Memory, attention, ACT-R model Cognitive AI,
cognitive executive control cognitive
structure simulations, training
systems
Models in working memory

1. Baddeley and Hitch Model (1974)

This is one of the most influential models of working memory.

Components

1. Central Executive
o Acts as the control system or “attention manager.”
o Directs attention, coordinates information from subsystems, and manages
problem-solving and decision-making.
o Functions: Task switching, inhibition of irrelevant information, planning.
2. Phonological Loop
o Responsible for verbal and auditory information.
o Subcomponents:
 Phonological Store: Holds spoken words for 1–2 seconds.
 Articulatory Rehearsal Process: Repeats information to keep it
active.
o Example: Remembering a phone number by repeating it silently.

3. Visuospatial Sketchpad
 Handles visual and spatial information.
o
Helps in mental imagery, navigation, and spatial reasoning.
o
Example: Visualizing a route to your friend’s house or manipulating objects
o
mentally.
4. Episodic Buffer (added in 2000)
o Integrates information from phonological loop, visuospatial sketchpad, and
long-term memory into a coherent, time-sequenced “episode.”
o Supports conscious awareness and complex cognitive tasks.
o Example: Remembering a story while integrating visual images, dialogue,
and your prior knowledge.

Diagram (conceptual):

Strengths:

 Explains multitasking and different cognitive tasks separately.

 Supported by neuroscience evidence (e.g., brain regions linked to verbal vs. spatial
memory).
Limitations:

 The central executive is vaguely defined; exact mechanisms are unclear.

2. Cowan’s Embedded-Processes Model (1999)

Cowan proposed a model focusing on attention and the capacity of working memory.

Key Points

1. Working memory is not separate from long-term memory; instead, it is an

activated portion of long-term memory.
2. Focus of Attention:
o A small subset of activated memory items (typically 3–5 chunks) is held in the
focus of attention for immediate use.
3. Working memory capacity is limited by attentional resources rather than storage
space.

Example:

 When mentally calculating 27 × 4, you temporarily activate numbers and intermediate

results from long-term memory and hold them in the focus of attention.

Strengths:

 Explains the close interaction between working memory and long-term memory.
 Emphasizes the role of attention in cognitive tasks.

Limitations:

 Less detailed about the mechanisms of rehearsal and manipulation of information

compared to Baddeley’s model.

3. Atkinson & Shiffrin Model (1968) – Multi-Store Model

Although more of a memory model than working memory, it provides a framework for
understanding short-term memory:

1. Sensory Memory: Briefly holds sensory input (milliseconds to seconds).

2. Short-Term Memory (STM): Temporarily holds information for processing (like
working memory).
3. Long-Term Memory (LTM): Permanent storage of knowledge.

Working Memory Role:

  STM in this model corresponds to working memory, which interacts with LTM for
encoding, retrieval, and manipulation of information.

Example:

 Reading a sentence requires holding words in STM (working memory) while

integrating meaning from LTM.

4. Oberauer’s Model (2002) – Three-State Model

Oberauer proposed that working memory has three functional states:

1. Activated Long-Term Memory (aLTM): All information currently activated but not
in immediate focus.
2. Region of Direct Access: Smaller set of items that can be readily accessed for
cognitive tasks.
3. Focus of Attention: Single item or chunk currently being processed.

Example:

 Solving a math problem: numbers are activated in aLTM, recent steps are in the
region of direct access, and the current step is in the focus of attention.

Strengths:

 Explains rapid access to some items versus slower retrieval of others.

 Aligns with neurocognitive evidence of memory and attention.

Summary Table of Models

Model Key Concept Components Strength

Multi-
Baddeley & Central Executive, Phonological Loop, Explains multitasking and
component
Hitch Visuospatial Sketchpad, Episodic Buffer modality-specific memory
system

Attention- Highlights attention and

Cowan Activated LTM + Focus of Attention
focused capacity limits

Atkinson & Classic framework

Multi-store Sensory Memory, STM, LTM
Shiffrin connecting STM and LTM

Three functional aLTM, Region of Direct Access, Focus Explains fast vs. slow access
Oberauer
states of Attention to memory items

Applications of Working Memory Models

 Education: Tailoring teaching methods based on memory capacity.
 AI & Cognitive Computing: Designing systems that mimic human short-term
information processing.
 Clinical Psychology: Understanding memory deficits in ADHD, dementia, or
traumatic brain injury.
 Human-Computer Interaction: Creating interfaces that reduce working memory
load.
1. Decay Process

Definition:

The decay process refers to the gradual loss of information over time when it is not
actively maintained or rehearsed in working memory.

Key Points:

 Memory traces fade naturally if they are not encoded into long-term memory.
 The rate of decay depends on attention, frequency of use, and significance of the
information.
 Decay is especially significant in short-term or working memory.

Example:

 Forgetting a phone number after a few minutes because it was never written down or
repeated.
 In a classroom, students may forget details of a lecture if not reviewed or applied.

Implications for Cognitive Models:

 Cognitive models must account for memory decay when simulating short-term
retention and decision-making over time.
 Helps design rehearsal strategies or reminders in learning systems.

2. Recall Process

Definition:

The recall process is the retrieval of stored information from memory when needed for
cognition, reasoning, or action.

Key Points:

 Can be free recall (retrieving without cues) or cued recall (retrieving with hints).
 Accuracy depends on the strength of memory traces, encoding quality, and
interference from other memories.
 Recall interacts closely with attention and working memory.

Example:

 Answering a question in an exam from previously studied material.

 A doctor remembering patient symptoms from prior cases to make a diagnosis.

Implications for Cognitive Models:

  Models must simulate how memories are retrieved and how retrieval failures or
errors occur.
 Important for AI systems simulating human-like memory or for educational
cognitive models.

3. Rehearsal

Definition:

Rehearsal is the mental repetition of information to maintain it in working memory or

transfer it into long-term memory.

Key Points:

 Maintenance Rehearsal: Repeating information to keep it temporarily active.

 Elaborative Rehearsal: Linking new information with existing knowledge to
enhance long-term storage.
 Reduces decay by actively refreshing memory traces.

Example:

 Repeating a phone number several times until dialed (maintenance rehearsal).

 Connecting a new concept in math with previously learned formulas (elaborative
rehearsal).

Implications for Cognitive Models:

 Models must include rehearsal mechanisms to simulate memory retention.

 Critical in AI systems for learning and adaptive behavior.

4. Sufficiency of a Cognitive Model

Definition:

The sufficiency of a cognitive model refers to how adequately the model explains and
predicts human cognitive processes. A model is considered sufficient if it captures essential
aspects of cognition without oversimplification.

Key Points:

 Should account for perception, attention, memory, reasoning, decision-making,

and learning.
 Must predict behavior accurately across various tasks and situations.
 Balances complexity and usability: Too simple → insufficient; too complex →
impractical.
Example:

 ACT-R cognitive architecture is considered sufficient because it models memory,

attention, learning, and problem-solving together.
 A model that only simulates short-term memory without attention or decision-making
would be insufficient.

Implications:

 Sufficiency ensures that the model can be used for AI, human-computer
interaction, and cognitive research.
 Guides researchers in validating and refining models.

Summary Table

Concept Definition Example Cognitive Modeling

Implication
Decay Process Gradual loss of Forgetting a phone Models must simulate
information over number after a few memory fading and short-
time minutes term retention
Recall Process Retrieval of stored Answering an exam Models must include
information question from memory retrieval
memory mechanisms
Rehearsal Repetition to Repeating a phone Important for maintaining
maintain or encode number or linking new working memory and long-
memory concepts term storage
Sufficiency of a Adequacy in ACT-R architecture Ensures model is
Cognitive Model explaining and modeling memory and comprehensive, accurate,
predicting cognition learning and practical
Verisimilitude Versus Truth
1. Truth vs. Verisimilitude

In philosophy and cognitive science, truth and verisimilitude are related but different
concepts:

Truth

 Refers to absolute correctness or conformity with reality.

 A statement is true if it accurately reflects facts or reality.
 Truth is binary: either a statement is true or it is false.

Example 1:

 Statement: "Water boils at 100°C at sea level."

 This is true, because it accurately reflects the physical fact under the stated
conditions.

Example 2:

 Statement: "The Earth revolves around the Sun."

 True, as it aligns perfectly with astronomical facts.

Verisimilitude

 Refers to closeness to truth, or the appearance of truth.

 A statement may not be completely true, but it can still be plausible, credible, or
likely.
 Useful in reasoning under uncertainty or incomplete knowledge.

Example 1:

 Statement: "It will probably rain tomorrow."

 This is not strictly true or false yet, but it has verisimilitude if meteorological data
support it.

Example 2:

 Statement: "Dragons might have existed in ancient times."

 Not true according to current scientific knowledge, but in folklore or literature, it has
verisimilitude (appears credible in context).

Summary:

 Truth = objective correctness.

 Verisimilitude = plausible or believable closeness to truth.
Relationship Between Truth and Verisimilitude

 Verisimilitude is a measure of how close an idea, theory, or statement is to the

truth.
 In cognitive systems and AI:
o Truth is rarely fully accessible due to incomplete or uncertain data.
o Verisimilitude guides decisions, predictions, and reasoning based on the most
plausible available information.

Example in AI:

 A medical diagnosis system cannot always give the absolute truth (disease present or
absent), but it assigns probabilities and suggests the most verisimilar diagnosis,
which helps doctors make decisions.

2. Attributes Used in a Cognitive System

A cognitive system is designed to mimic human cognition. Various attributes define its
capabilities and effectiveness:

Attribute Explanation
Perception Ability to sense and interpret data from the environment (vision,
speech, text).
Attention Focus on relevant inputs and ignore irrelevant ones.
Memory Storing and recalling knowledge, both short-term (working
memory) and long-term.
Learning Ability to improve performance from experience and adapt to new
situations.
Reasoning Drawing conclusions, making decisions, or solving problems based
on knowledge.
Knowledge Representation Structuring data in a way the system can manipulate (semantic
networks, frames).
Language Understanding Comprehending, generating, and interacting through natural
language.
Planning & Decision-Making Selecting actions based on goals and predicted outcomes.
Self-awareness / Meta- Monitoring its own processes to adjust strategies or improve
cognition efficiency.
Social Cognition Interpreting and responding to human emotions, intentions, and
social cues.

Example of Cognitive System Application:

 IBM Watson: Uses perception (text/speech input), reasoning (answers questions),

learning (improves with new data), and language understanding (natural language
queries).
Nature of Explanations in cognitive computing
1. Introduction

In cognitive computing, systems are designed to simulate human reasoning. Unlike

traditional rule-based systems, cognitive systems interact with humans, learn from data, and
provide explanations for their decisions or actions.

Explanations are crucial because they make AI systems transparent, trustworthy, and
interpretable. Humans need to understand why a system arrived at a particular
conclusion to validate or act on it.

2. Nature of Explanations

The nature of explanations in cognitive computing can be characterized by several key

features:

a) Contextual

 Explanations depend on the context of the problem or user query.

 A cognitive system considers environmental factors, prior knowledge, and user
expectations.
 Example: In a healthcare AI system, the explanation for a diagnosis includes patient
history, symptoms, and test results.

b) Probabilistic / Approximate

 Cognitive systems often deal with uncertain or incomplete information.

 Explanations are probabilistic rather than deterministic, reflecting the degree of
confidence in the decision.
 Example: "Based on your symptoms, there is a 75% likelihood of influenza."

c) Interactive / Iterative

 Explanations are user-centric and can be refined based on follow-up questions.

 Users can query the system, and it can provide deeper insights.
 Example: A financial AI can explain why a loan application was rejected, and the
user can ask for more details on specific risk factors.

d) Multi-level / Hierarchical
  Explanations can be provided at different levels of detail:
1. High-level summary for quick understanding.
2. Intermediate level showing reasoning steps.
3. Detailed level including raw data and algorithms.
 Example: In a recommendation system:

o High-level: "You may like this movie because you enjoy action films."
o Detailed: "Your past ratings for action movies are 4.5/5 on average, and users
with similar patterns rated this movie 4.7/5."

e) Transparent / Traceable

 Explanations should allow users to trace decisions back to data, models, or

reasoning steps.
 Transparency builds trust and supports auditability.
 Example: In autonomous vehicles, the system can explain: "Braked because an object
was detected in the lane at 15 meters distance."

f) Human-centric

 Explanations must be interpretable in human terms, not just technical jargon.

 Cognitive systems often use natural language explanations, visualizations, or
analogies.
 Example: A chatbot explaining market trends might say: "The stock price fell
because demand decreased and supply increased."

g) Adaptive

 Cognitive systems adapt explanations based on user knowledge.

 Experts get detailed technical reasoning, while novices get simplified reasoning.
 Example: In medical AI:
o Expert doctor: "The lesion shows hyperintensity in T2-weighted MRI,
consistent with glioma."
o Patient: "The scan shows an unusual growth that could be a type of brain
tumor."

3. Importance of Explanations

1. Trust & Acceptance: Users trust AI systems more when they understand decisions.
2. Decision Support: Explanations help humans make informed decisions alongside
AI.
 3. Debugging & Improvement: Developers can identify errors in reasoning or data.
4. Ethical & Regulatory Compliance: Required in areas like healthcare, finance, and
law.

4. Techniques Used for Explanations in Cognitive Systems

 Rule-based reasoning explanations: Trace decisions through logical rules.

 Model-agnostic techniques: LIME, SHAP, which explain predictions of complex
models.
 Natural language generation (NLG): Converts decision logic into human-readable
explanations.
 Visualization: Graphs, charts, and diagrams to show patterns or decision paths.

Summary:
The nature of explanations in cognitive computing is contextual, probabilistic, human-centric,
adaptive, interactive, and transparent. They are essential for making AI decisions interpretable,
trustworthy, and actionable.