Cognitive Models

Introduction

Cognitive Models predict user thoughts and reactions when interacting with systems. These models describe how users work to interact with interfaces, from high-level goals and problem-solving down to low-level motor actions (such as keystrokes).

Goals of Cognitive Modeling:

• Understand mental processes of users when using systems
• Predict time required to complete tasks
• Identify potential interaction difficulties
• Design interfaces that match human cognitive capabilities

---

Types of Cognitive Models

1. MHP (Model Human Processor)

Model that treats humans as information processing systems

2. GOMS (Goals, Operators, Methods, Selection)

Hierarchical model that breaks down user tasks into components

3. KLM (Keystroke-Level Model)

Model for analyzing unit tasks with low-level detail

---

1. Model Human Processor (MHP)

Developed by Card, Moran, & Newell in the 1980s. This model treats humans as information processing systems with components working in sequence.

MHP Components

Basic Structure:

• Memories and processors working in specific order
• Clear principles of operation
• Discrete stages with characteristic timing for each

Subsystems in MHP

1. Perceptual Subsystem

• Function: Sensors + sensory image storage (visual & audio)
• Process: Encodes sensory input symbolically
• Processing Time: ~100ms for visual recognition
• Capacity: Limited to information that can be processed simultaneously

2. Cognitive Subsystem

• Function: Receives symbolic codes into working memory
• Process: Combines with long-term memory (LTM) to decide actions
• Processing Time: ~70ms per cognitive cycle
• Capacity: Working memory limited (~7±2 items)

3. Motor Subsystem

• Function: Executes actions stored in LTM based on working memory commands
• Process: Controls physical movement and motor responses
• Processing Time: ~70ms for movement initiation
• Variability: Depends on movement complexity

Total Time Calculation

Total Processing Time = Σ(individual stage times)

Example Applications:

• Calculate time needed to read and respond to error messages
• Predict user reaction time to visual changes
• Design optimal timing for animations and transitions

---

2. GOMS (Goals, Operators, Methods, Selection)

Hierarchical model that breaks user tasks into components that can be systematically analyzed.

GOMS Components

Goals

• Definition: End states that users want to achieve
• Structure: Can be broken into smaller sub-goals
• Examples: "Close application window", "Save document", "Send email"

Operators

• Definition: Basic actions users can perform
• Types: Physical (mouse click, key press) and mental (remember, decide)
• Examples: Press key, drag mouse, type text

Methods

• Definition: Sequences of operators to achieve specific goals
• Characteristics: Organized procedural steps
• Flexibility: One goal can be achieved through various methods

Selection Rules

• Definition: Criteria for choosing methods based on context
• Factors: User experience, situation, preferences
• Implementation: If-then logic for method selection

Example GOMS Analysis: Closing a Window

GOAL: CLOSE-WINDOW
  MENU-METHOD:
    MOVE-MOUSE-TO-FILE-MENU
    PULL-DOWN-FILE-MENU
    CLICK-CLOSE-OPTION

  CTRL-W-METHOD:
    PRESS-CONTROL-W-KEYS

  ALT-F4-METHOD:
    PRESS-ALT-F4-KEYS

Selection Rules:

1. Use menu method by default (for novice users)
2. If in a game, use Alt-F4 (for speed)
3. If typing, use Ctrl-W (keyboard efficiency)

Comparative Analysis: PC DOS vs. Macintosh

Cognitive Analysis:

• PC DOS: Requires more memory (remembering command syntax)
• Mac: More intuitive but requires more physical steps

Benefits of GOMS Analysis

For Designers:

• Measure expert user performance
• Estimate short-term memory load
• Test method selection accuracy
• Compare design alternatives

For Developers:

• Optimize workflows
• Identify cognitive bottlenecks
• Design effective shortcuts

---

3. KLM (Keystroke-Level Model)

Model for unit tasks (<20 seconds) after decomposition using methods like GOMS. This model focuses on detailed low-level analysis of user interactions.

Operators and Standard Times

Rules for Mental Operator (M) Placement

When to Insert M:

1. Before sequences of physical operators directed to different goals
2. Before selecting commands from menus
3. Before operations requiring special concentration

When NOT to Insert M:

1. Between operators in the same group
2. After P operator that already includes thinking time
3. In sequences that are already automatic

Example 1: Edit One Character

Operator Sequence:
H[mouse]               ; move hand to mouse
P B[LEFT]              ; point & click to select bad character
H[keyboard]            ; return to keyboard
M K[DELETE]            ; think + delete
K[char]                ; type correction
H[mouse] P B[LEFT]     ; reposition cursor & click

Time Calculation:
TEXECUTE = 2K + 2B + P + 3H + 2M
        = 2(0.28) + 2(0.20) + 1.10 + 3(0.40) + 2(1.35)
        = 0.56 + 0.40 + 1.10 + 1.20 + 2.70
        = 5.96 seconds

Example 2: Minimize Window

GOAL: MINIMIZE-WINDOW

MINIMIZE-BUTTON-METHOD:
  P[to minimize button] = 1.1
  B[LEFT down/up] = 0.1 + 0.1
  M = 1.35
  Total ≈ 2.65 seconds

KEYBOARD-SHORTCUT-METHOD:
  H[to keyboard] = 0.40
  M = 1.35
  K[Windows+M] = 0.28
  Total ≈ 2.03 seconds

RIGHT-CLICK-MENU-METHOD:
  P[right-click taskbar] = 1.1
  B[RIGHT] = 0.20
  M = 1.35
  P[minimize all] = 1.1
  B[LEFT] = 0.20
  Total ≈ 3.95 seconds

Practical Applications of KLM

Design Evaluation:

• Compare efficiency of different input methods
• Optimize interface layout
• Identify time-consuming interactions

Prototype Validation:

• Predict task completion time before implementation
• Measure performance improvement from design changes
• Set realistic performance targets

---

Limitations of Cognitive Models

MHP Limitations

• Linearity Assumption: Not all processing occurs sequentially
• Individual Variability: Differences in cognitive abilities between users
• Situational Context: Doesn't consider environmental factors

GOMS Limitations

• Expert Users Only: Doesn't predict learning or errors
• Routine Tasks: Less effective for creative or exploratory tasks
• Real Complexity: Difficult to handle very complex interactions

KLM Limitations

• Micro Focus: Only analyzes low-level tasks
• Ignores Errors: Doesn't account for error recovery
• Limited Context: Doesn't consider global cognitive load

---

Applications in HCI Design

Design Phase

Requirements Analysis:

• Use GOMS to understand user workflows
• Identify main goals and sub-goals
• Map alternative methods for each goal

Prototype Evaluation:

• Apply KLM to measure interaction efficiency
• Compare execution time of different design alternatives
• Optimize based on model predictions

Implementation Phase

Performance Optimization:

• Use MHP to determine system response timing
• Adjust animation speed to perceptual capabilities
• Design feedback appropriate to cognitive cycles

Testing and Validation:

• Compare model predictions with user testing results
• Calibrate model parameters based on empirical data
• Iterate design based on discovered discrepancies

---

Summary

Cognitive models provide a scientific framework for understanding and predicting human-computer interaction. Each model has strengths and limitations that must be understood for effective application.

Key Principles

1. Model Complementarity

• Use combination of models for comprehensive analysis
• MHP for fundamental understanding, GOMS for task structure, KLM for details
• Match model choice to development stage and analysis goals

2. Empirical Validation

• Models are predictions that must be validated with real user data
• Use models as starting points, not absolute truth
• Iterate and improve models based on testing findings

3. Application Context

• Consider target user characteristics (expert vs. novice)
• Adapt to application domain and task types
• Integrate with other HCI evaluation methods

Implementation Benefits

For Design Teams:

• Performance Prediction: Estimate time and efficiency before implementation
• Design Optimization: Identify bottlenecks and improvement areas
• Decision Validation: Scientific basis for design choices

For Organizations:

• Development Efficiency: Early-stage problem detection
• Better ROI: Targeted investment in meaningful improvements
• Process Standardization: Consistent methodology for design evaluation

Cognitive models, when properly applied, can significantly improve the quality and effectiveness of user interface design.