Skip to main content Skip to footer site map
David Joyner, Math Department
math_007.

Learning complex tasks

Factors Affecting Complexity

Analysis is often an incredibly complex task, in that it involves extremely complicated judgment, planning, and decision making, and requires extensive expert-level knowledge and skill. They are tasks that require deep expertise and highly focused practice for successful performance. Complex tasks involve abstractions rather than concrete phenomena. They usually involve multiple interacting causes or sources of variation that affect outcomes. They involve non-linear processes that are dynamic, continuous, and have many simultaneously "moving parts." They are characterized by uncertainty and ambiguity. Some of the properties of tasks that contribute to complexity are given in the Table below.

Property Description
Abstract Physical phenomena or causation are not readily visible
Multivariate Many variables underlie outcomes.
Interactive Changes in one variable may affect several others. Processes are co-dependent.
Continuous Physical phenomena and their effects are described as values along continua, rather than as discrete properties.
Non-Linear Relations among variables are not simple straight-line functions
Dynamic The process of variation is of interest, rather than end-state
Simultaneous Systemic variation is coincident rather than serial.
Conditional Outcomes are highly dependent on boundary conditions and context.
Uncertainty & Ambiguity The same outcome may arise from different combinations of inputs.

Design guidelines for training and performance aiding systems to support complex tasks.

Complexity and Competence

Training for incredibly complex tasks needs to
  1. teach principles underlying abstraction,
  2. help learners cope with multiple sources of variability in the problem space,
  3. provide aids that reduce memory load,
  4. provide strategies for dealing with uncertainty and ambiguity, and
  5. provide training in problem solving, judgment, and decision making.
Some design guidelines for training and performance aiding systems for complex tasks are given in the Table below.
Property Design Guideline
Abstract Develop visualizations that explain underlying physical phenomena and causation
Multivariate Develop simulation-based/physics-based problem space in which effects of variation can be explored
Continuous Provide for continuous variation, not a succession of states
Non-Linear Explore the non-linearity: Concentrate on inflection points, minima, maxima, zero-crossings, asymptotes
Dynamic Task environment must include dynamic complexity – Scenarios must present continuous evolution, not a succession of static states
Interactive Systematically hold some variables constant while exploring variation. Use no more than three-way interactions for problem cases
Simultaneous Develop mental models for simultaneity as underlying interaction, not serial causation
Conditional Provide highly contextualized environment that is capable of supporting practice in high-difficulty real-world environments
Uncertainty & Ambiguity Develop methods / procedures for resolving uncertainty / ambiguity. Task environment must properly replicate these effects. Develop test scenarios that exploit uncertainty.

Taken from sections 4.4B and 4.4C (pages 135-136) of Collaboration in the National Security Arena: Myths and Reality - What Science and Experience Can Contribute to its Success, June 2009. Compiled by: Nancy Chesser (DTI), Nancy.Chesser@js.pentagon.mil
wdj@usna.edu
go to Top