Innovative Learning

Model-Centered Instruction/Design Layers (Andrew Gibbons)

Model-Centered Instruction (MCI) is a set of principles to guide instructional designers in selecting and arranging design constructs, so it is appropriately called a design theory. It favors designs that originate with and maintain the priority of models as the central design

Background: A Layered View of Design—MCI is closely tied to a layered view of designs. This view assumes that a designer organizes constructs within several somewhat independent layers characteristic of instructional designs: the model/content layer, the strategy layer, the control layer, the message layer, the representation layer, the media-logic layer, and the management layer. The designer selects and organizes structures within each layer in the process of forming a design. The designer also aligns the structures within layers with those of other layers to create a vertical modularity in the design that improves its manufacturability, maintainability, and the reusability of designed elements. A design layer is typified by: characteristic design goals, building-block constructs, design processes, design expression and construction tools, and principles to guide the arrangement of structures. Over time, a layer becomes associated with specialized skill sets, publications, and a design culture. Instructional theories provide principles to guide design within one or more of these layers, but no theory provides guidelines for all of them, suggesting to designers the wisdom of subscribing to multiple local theories of design rather than a single monolithic theory.

MCI Theory: Model-Centered Instruction, as any design theory, can be described in terms of the prescriptive principles it expresses for each of these layers.

Content: The content of instruction should be perceived in terms of models of three types: (1) models of environments, (2) models of cause-effect systems (natural or manufactured), and (3) models of human performance. Together these constitute the elements necessary for performance and therefore for learning. Content should be expressed relative to the full model structure rather than simply as facts, topics, or lists of tasks.

Strategy: The strategy of instruction should be perceived in terms of problems. A problem is defined as any self-posed or instructor/designer-posed task or set of tasks formed into structures called “work models” (Gibbons, et al., 1995). These are essentially scoped performances within the environment, acting on systems, exhibiting expert performance. Problems may be presented as worked examples or as examples to be worked by the learner. During problem solution instructional augmentations of several kinds may be offered or requested. Dynamic adjustment of work model scope is an important strategic variable.

Control: Control (initiative) assignment should represent a balance between learner and instructor/designer initiatives calculated to maximize learner momentum, engagement, efficient guidance, and learner self-direction and self-evaluation. Instructional controls (manipulative) should allow the learner maximum ability to interact with the model and the instructional strategy’s management.

Message: Contributions to the message arise from multiple sources which may be architecturally modularized: (1) from the workings of the model, (2) from the instructional strategy, (3) from the controls management, (4) from external informational resources, and (5) from tools supplied to support problem solving. The merging of these into a coherent, organized, and synchronized message requires some kind of message or display management function.

Representation: MCI makes no limiting assumptions about the representation of the message. Especially with respect to model representation, it anticipates a broad spectrum of possibilities—from externalized simulation models to verbal “snapshots” and other symbolics that call up and make use of models learners already possess in memory.

Medial-Logic: MCI makes no assumptions regarding the use of media. Its goal is to achieve expressions that are transportable across media. The selection of the model and the problem as central design constructs assist in this goal.

Management: MCI makes no assumption about the data recorded and used to drive instructional strategy except to the extent that it must parallel the model’s expression of the content and align also with the chosen units of instructional strategy.


When the designer enters design from the model/content layer, the priority of concerns follows this order:

(1) What is the appropriate cause-effect model (or system) the learner should interact with?

(2) What is the appropriate level of denaturing (reduction in fidelity and granularity) of models for a given learner?

(3) What sequence or set of problems should the learner solve as a lens into or a mask on this model?

(4) What resources and tools should be available as solving takes place?

(5) What additional instructional augmentations should be supplied to support the solving of the problem?

Designers can (and do) enter design at any layer, placing highest priority on one of them. Design decisions made within the priority layer, however, then constrain decisions within the remaining layers and often either create or destroy other layers and sub-layers of the design. This principle leads to important insights into the order of instructional design activities and thus layers provide a basis for generating and ordering design processes dynamically.

An analysis approach called the Model-Centered Analysis Process (MCAP) identifies the elements of all three model types and relates them directly to problems. This automatically unites the specification of the learning environments, instructional functionalities, surface dramatics, and logical structures (if a computer is to be involved, which is not assumed).


A model-centered design is centered around the model(s) selected by the designer. This is often a difficult and subtle choice. It is easy, for example, for a designer to mistakenly provide an interactive panel simulation for chemical analysis equipment when what is needed is observation and interaction with an expert model of interpreting the outcome of chemical tests. The panel model can become the center of the designer’s attention because it is concrete and programmable, shifting attention away from the more important performance model that the learner would benefit from more.


The principles of model-centered instruction are:

  1. Experience: Learners should be given maximum opportunity to interact for learning purposes with one or more systems or models of systems of three types: environment, system, and/or expert performance. The terms model and simulation are not synonymous; models can be expressed in a variety of computer-based and non-computer-based forms.
  2. Problem solving: Interaction with systems or models should be focused by the solution of one or more carefully selected problems, expressed in terms of the model, with solutions being performed by the learner, by a peer, or by an expert.
  3. Denaturing: Models are necessarily denatured from the real by the medium in which they are expressed. Designers must select a level of denaturing matching the target learner’s existing knowledge and goals.
  4. Sequence: Problems should be arranged in a carefully constructed sequence for modeled solution or for active learner solution.
  5. Goal orientation: Problems selected should be appropriate for the attainment of specific instructional goals.
  6. Resourcing: The learner should be given problem solving information resources, materials, and tools within a solution environment (which may exist only in the learner’s mind) commensurate with instructional goals and existing levels of knowledge.
  7. Instructional augmentation: The learner should be given support during solving in the form of dynamic, specialized, designed instructional augmentations.


Duffin, J.W.  & Gibbons, A.S. (2001) Decompressing and Aligning the Structures of CBI Design.

Gibbons, A. S. (in press). Model-Centered Instruction. Journal of Structural Learning and Intelligent Systems.

Gibbons, A. S., Bunderson, C. V., Olsen, J. B., & Robertson, J. (1995) Work models: Still beyond instructional objectives. Machine-Mediated Learning, 5(3&4), 221-236.

Gibbons, A. S., & Fairweather, P. G. (1998) Computer-Based Instruction: Design and Development. Englewood Cliffs, NJ: Educational Technology Publications.