In order to understand appropriate use of technology in teaching, we need to understand which aspects of our curriculum are critical for which aspects of our future learning. For example, if general hand / eye coordination is transferable across tasks, do we need to agonise over detailed task-specific simulations for each task we want to learn? And if hand / eye coordination is transferable, do we need a few authentic, high fidelity simulations, or a broader range of generic simulations.
Let’s take a specific example of a visuo-motor task simulation using stereoscopic 3D vision and haptic (active touch) feedback from a hand-held probe. If haptic feedback is only helpful with the level of fidelity provided by a stereoscopic visual display, this implies that it is critical for the visuo-haptic feedback to be ecologically valid and temporally faithful, and for the task model to be authentic. If ecological validity and authenticity are not critical, the question must be asked as to which specific aspects of the simulation are important (rather than just “cool”)?
If active exploration of visuo-haptic-motor space versus passive presentation of sensory information is critical, then would it be enough to provide a 2D visual display and allow active exploration through joystick control? Does it depend on whether you are learning how to perform the specific task (where differential haptic feedback might be important) versus learning how to conceptualise the specific task (where joystick-controlled exploration might be enough)? Does the “analogue” continuous control provided by the joystick teach something different from the digital discrete-step control provided by keyboard controls? Would haptic feedback through specific sequences of pre-determined tasks (selected for cue salience) provide better learning outcomes than free exploration of the model space (which may not result in the student exposing themselves to relevant comparisons from which they can extract salient task cues)?
Research into haptic feedback is a perfectly reasonable undertaking, but is there enough evidence to suggest that one should invest in 3D VR models and haptic probes for real training and education situations? Not only does the haptic feedback need to have a demonstratable impact on learning outcomes, but the learning outcomes must be critical to the goals of teaching and financially supportable compared with other teaching strategies. For high fidelity visuo-haptic simulations to be financially viable, new task models would need to be able to be developed more cheaply and quickly than the active life of the simulation hardware, the haptic feedback itself would need to have made a significant contribution to the specific task learning outcomes, and the learning outcomes would have to be achieved more cheaply than by other available methods.
How do you test the learning outcomes? Is it in terms of specific skill acquisition? Should task-specific training also be generalisable to other similar tasks? In order to address these questions, there needs to be some theoretical understanding of the nature of the task and a clear articulation of task-critical features. Some differential learning situations (in presumed order of ecological validity) to think about for example in a surgical scenario are:
1) training on cadavers / animal models
2) training on 3D models with haptic feedback
3) training on 2D models with interactive exploration (joystick / keyboard control)
4) “yoked control” observation of surgery (real and simulated)
In considering differential learning outcomes, what framework of learning do we assume? Do we have measures which can provide critical data to distinguish between hypotheses? What would we consider to be evidence for improved outcomes between, for example, task 2 and task 3? Is task 1 the appropriate control condition, is it a test condition, and / or is actual surgical performance the only valid “test” of learning? Without a strong model for how haptic feedback enhances skill acquisition, it is difficult to provide a strong justification for the significant financial outlay in introducing VR + haptic feedback training solutions.
How do the issues raised for this specific simulation example pan out for other teaching paradigms? I contend that we need to be able to identify specific teaching and / or learning objectives and to take a position about how our practice brings about the achievement of those objectives. This necessarily entails a conceptualisation of what occurs during teaching and learning. It also requires some position on short-term versus long-term learning, specific skills versus task-specific knowledge versus discipline-area knowledge, and the metacognitive processing around these things.
Do we take a position that declarative knowledge and procedural knowledge are acquired differently? If so, what primary learning outcomes are desired – skill based or discipline based? How do we conceptualise the relationship between theoretical knowledge and practical skills deriving from that technical knowledge? I contend that without taking a position on these questions, we cannot make informed decisions about educational design. Furthermore, for practical skills, we can break down requirements with respect to domain knowledge about a domain versus metaknowledge and insight around aquiring domain-specific practical skills. Does evidence from the perceptual-motor learning area suggest different approaches to teaching dependent on the requirement for repetitive accuracy of well-defined motor skills in a well-defined environment, versus creative perceptuomotor response and adaptability for a defined outcome in a changeable environment? Do we need to incorporate redundancy in simulations to ensure exposure to multiple cues to increase the potential for adaptability in changeable environments or are we aiming to extract the essential, minimal cue set to perform a task to become extremely efficient? What are the implications for robustness, pattern recognition (synthesis / generalisation …) and rule extraction (analysis / abstraction …) of different forms of simulations?
(now also posted on my OLU blog)