Cognitive Task Analysis (OFM-COG)

The third example of workload assessment using task loading measures is interesting because the method was developed specifically for use in a maritime context to assess the workload implications of ship-borne automation systems. The approach is based upon a state-transition type of task analysis and this is used to develop an Operator Function Model (OFM). The OFM provides a representation of the operator functions, sub-functions, behaviours and information needs constructed using the language and methods of discrete control event modelling. The technique was first developed in the mid 1980s and has been used on a number of occasions to evaluate operator behaviour in a number of high risk work environments (e.g., Miller, 1985; Mitchell and Miller, 1986). Mitchell and Miller claim that OFMs provide a rich, dynamic, task analytic structure for defining job profiles, which can be employed as an alternative to traditional static task analysis methods (e.g., Mitchell, 1987).

In order to apply the approach to the evaluation of mental workload, Lee and Sanquist (2000) extended the mathematical base of the OFM model to include a description of cognitive transactions that impose cognitive load on the human operator. The resultant model – which has become known as OFM-COG, uses tables to compute cognitive load from the OFM task description. An example of one such table is provided in Figure 24.

Column 1 of the table provides a description of the main cognitive transactions which may be used in the construction of the cognitive Operator Function Model component. The table also provides a brief description of its function. Column 2 indicates the category of human information processing implicated in completing the transaction The HIP model used to drive the analysis is relatively simple and has only three information processing stages: Information Acquisition, Information Handling and Information Interpretation. Column 3 defines the cognitive resource requirement associated with the cognitive transaction. Lee and Sanquist propose nine resources types as shown in Figure 25:

Cognitive Agent

Task Category of IP HIP Resource requirement Input Select. Selecting what to pay

attention to next Acquisition

Selective attention Perceptual sensitivity Filter. Straining out unimportant data

streams Acquisition Selective attention Detect. Is something there? Acquisition Perceptual sensitivity _{Distributed attention} Search. Looking for something Acquisition Sustained attention

Perceptual sensitivity Identify. What is it, what is its name? Acquisition _Interpret Perceptual discrimination Working memory

Long-term memory Prepare Message. Prepare collection of

symbols for sending as a meaningful statement

Handling Response precision Queue to channel. Lining up a process for

near future performance Handling Working memory Processing strategies Code. Translate item from one for to

another Handling

Response precision Working memory Long-term memory Transmit. Move something from one

place to another Handling Response precision Store. Keep something for future use Handling Working memory _{Long-term memory}

Store in Buffer. Hold something

temporarily Handling

Working memory Processing strategies Compute. Figure out something logically

or mathematically answer a defined problem

Handling Working memory Processing strategies Edit. Arranging or correcting things

according to the rules Handling Selective attention Long-term memory Display. Show something that makes

sense Handling Response precision

Purge. Clear data _{Handling Selective attention} Reset. Getting ready for some different

action Handling Selective attention Response precision Count. Keep track of how many Handling

Interpretation

Sustained attention Working memory Control. Change an action according to a

plan Handling Interpretation Response precision Decide/Select. Choose a response to fit

the situation. Interpret

Long-term memory Processing strategies Plan. Matching resources in time to

expectations. Interpret Working memory Processing strategies Test. Are things what they should be? Interpretation

Perceptual sensitivity Working memory Long-term memory Interpret. What does this mean? Interpretation Long-term memory _{Sustained attention} Categorise. Define and name a group of

things Interpretation Long-term memory Perceptual sensitivity Adapt/Learn. Making and remembering

new responses to a learned situation. Interpretation Long-term memory Goal Image. A picture of the task well

done Interpretation Long-term memory Processing strategies Figure 24 Miller's Cognitive Task Transaction List and require IP resource

• Perceptual Sensitivity • Perceptual Discrimination • Selective Attention • Distributed Attention • Sustained Attention • Working Memory • Long-term Memory • Response Precision • Processing Strategy

Figure 25 The Nine Cognitive Resource Types

The presumed relationship between each information processing strategy and its associated resource demand is summarised in Figure 26.

Information processing

category Resource Demand

Information Acquisition

Perceptual sensitivity Perceptual discrimination Working memory

Response precision

Information Handling Selective attention Sustained attention Distributed attention

Information Interpretation Long-term memory Processing strategy

Figure 26 Resource demand for each presumed stage of information processing The OFM-COG workload evaluation is essentially a qualitative analysis, which uses expert judgement to explore the implications of task structure on mental workload. A table is constructed, similar to that required for a Failure Modes and Effects Analysis (FMEA) in which cognitive transactions are used to seed consideration of a variety of factors that will contribute to workload. For example, Figure 27 shows the results of an evaluation of the sub-function “Course Execution” which is deemed to consist of four activities: (a) Determine position, (b) Record position, (c) Monitor progress, and (d) coordinate with VTS and Pilots. This task description is shown in Column 1 of the table. Column 2 records the transaction type. These are obtained from the descriptions already provided in Figure 24. It should be noted that the term cognitive agent is used here in its broadest sense and can refer to transactions implemented by the crew member or automated systems. In column 3, the input channel is recorded. Location of position uses GPS data

column records the presence of factors that will influence efficient task performance.

Figure 27 OFM-COG Analysis of track keeping sub-function within ECDIS