The human limb can send and receive information through either force/torque or displacement/rotation. Correspondingly, an isometric device connects the human limb and machines through force/torque while an isotonic device does this through movement.
Isometric devices are also called pressure devices or force devices. Literally, the word isometric derives from the Greek "isos" meaning "same" or equal and "metric" meaning "measure" or in other words, constant length, or non moving. According to WebsterÌs Ninth New Collegiate Dictionary, isometric means "of, relating to, or being muscular contraction against resistance, without significant shortening of muscle fibres and with marked increase in muscle tone". By this definition, an isometric device is a device that senses force but does not perceptibly move.
Isotonic devices are also called displacement devices, free moving devices or unloaded devices. From the Greek, the word isotonic means equal "tonikos", or constant tension. According to WebsterÌs Ninth New Collegiate Dictionary, isotonic means "of, relating to, or being muscular contraction in the absence of significant resistance, with marked shortening of muscle fibres, and without great increase in muscle tone - compare isometric". An isotonic device should have zero or constant resistance. The mice that are used with most of today's computer systems are examples of isotonic devices.
Between the isometric (infinite resistance) and the isotonic (zero or constant resistance) are devices with varying resistance. When the deviceÌs resistive force increases with displacement, the device is elastic, or spring-loaded. When resistance increases with velocity of movement, the device is viscous. Similarly, when the resistance increases with acceleration, it is an inertial device. In practice, all devices have some inertia. However, the device's inertia is usually ignored when it is relatively small compared to the inertia of the human hand or when the initial resistance is relatively small compared to other forms of resistance (e.g. elastic).
Some authors also use the term "moving device" as a short form for free moving (isotonic) device. Other authors have used it for all devices that move ("anisometric"). Anisometric devices however in fact include both free moving (isotonic) devices and elastic, viscous or inertial devices.
The peculiarities of many real world applications may favour either isometric or isotonic devices. For example, implementation with one device might simply be less costly than the other at a particular phase of technology development. Alternatively, certain work environments may not allow free hand movements due to physical workspace constraints or motion noise, such as in vehicles or aircraft. Leaving those special cases aside, however, the general performance differences between isometric and isotonic devices are of both theoretical and practical importance to human factors researchers, especially as they relate to modern multi-DOF controllers.
2.2.2 The Literature on isometric versus isotonic devices
Early research comparing isometric devices with isotonic devices, in the context of 1 or 2 DOF manual tracking, is well reviewed in Poulton (1974) . Poulton's hypothesis was that an isometric device ('pressure control' in his terminology) is in general advantageous whenever time is short, but disadvantageous when slow, accurate positioning is required. According to him, an isometric device has no travel time, which should make it quicker to control, but it can not be adjusted very accurately because it does not provide the human operator with any displacement cues proportional to its output. In contrast, an isotonic device or an elastic device ("moving control" in his terminology) does provide the displacement cue for accurate control .
Contradictory to Poulton's view, many other researchers, including Gibbs (1954) , Burke and Gibbs (1965) , argued that an isometric device should in fact provide stronger "proprioceptive discharge" and therefore should produce better performance for tracking tasks. Based on his experiments on manual tracking, Gibbs went on to advocate a "closed-loop" theory of motor control, since isometric controllers were believed to give more feedback to support closed-loop behaviour. GibbsÌ work has been influential in the motor control literature. For example, Keele (1986) cited Gibbs and promoted "the better quality and greater rapidity of kinaesthetic information in isometric muscle contractions as opposed to isotonic contractions".
Note that both views, represented by Poulton and Gibbs respectively, emphasised the importance of feedback. What they disagreed on was which device provides stronger feedback. Gibbs believed that an isometric device should give stronger feedback due to the stronger "proprioceptive discharge" since force is being used. Poulton, on the other hand, believed that anisometric devices give stronger feedback due to the "movement cue".
Poulton (1974) compiled a comprehensive list of studies that covered works from 1943 to 1966. Out of 17 investigations that he cited, 12 strongly favoured pressure control, two slightly favoured pressure control and only three slightly favoured anisometric (isotonic or elastic) controls. These studies were conducted under various conditions, ranging from rate control to position control, from high frequency tracking to slow ramp tracking, from compensatory to pursuit displays. Other reviews, such as Boff and Lincoln (1988, section 12.421), also give similar conclusions that isometric joysticks yield better performance (e.g. smaller tracking error).
In his speculations upon reasons for this contradiction to his hypothesis, Poulton pointed out that most of the studies used the balanced treatment (within-subjects) experiment design. He has been strongly against this type of experimental design in many of his publications (Poulton, 1966, 1969, 1973, 1974, 1989) . With a within-subjects design, he argued, the actual skill transfer from one condition to another might not be symmetrical, even when subjectsÌ exposures to the two conditions are equalised. In particular, for the case of isometric vs. isotonic control, the skill transfer might favour the isometric control. Poulton also noted that isometric devices are always spring centred while isotonic controls are not and he thus suspected that it might be the spring centring that caused the performance difference. Poulton (1974) concluded that in order to reach a definitive verdict between isometric and anisometric devices more experimental research was needed. Unfortunately, no further studies that explicitly followed Poulton's analysis have been found.
Notterman and Tufano (1980) took Gibbs' belief in the superiority of isometric kinaesthetic information and tested the so-called inflow-outflow debate in human motor control. Inflow theory proposes that human motor action fundamentally relies on feedback, the information flowing into the central nervous system (CNS) from the periphery. In contrast, outflow theory proposes that human motor control is primarily a result of executing motor commands flowing out of the CNS to the peripheral motor organs. On the basis of Gibbs' conclusion that isometric devices should give stronger feedback than isotonic devices, Notterman and Tufano argued that the relative human performance with an isometric device versus an isotonic device would be an indicator of the validity of inflow versus outflow theory. If superior performance were to be found with isometric devices, implying stronger feedback does improve human motor performance, inflow theory would be supported. On the other hand, if superior performance with isotonic devices were to be found, implying that human motor performance is actually better without or with less proprioceptive feedback, outflow theory would be supported. What Notterman and Tufano actually found was more complicated: (1) the isometric condition was better for randomly moving targets (0.33 Hz Gaussian noise) while the isotonic condition was better for predictably moving target (0.5 Hz sine waves). (2) the isotonic stick was better than an elastic stick at the beginning of training but worse by the end of training. They concluded that the inflow and outflow dispute was overly simplified. "Subjects profit from whatever exteroceptive and proprioceptive cues are available and efficacious and they organise their behaviour accordingly". Since Gibbs' notion of isometric superiority in proprioceptive feedback is questionable in any case, Notterman and Tufano's study did not actually have a solid basis for testing the inflow-outflow debate. Jones and Hunter (1990) conducted a systematic study on elastic resistance ranging from isotonic to isometric in a step tracking experiment. The major findings of their study confirmed what many early researchers had believed: stiffer devices can be used to generate faster responses, as indicated by (1) shorter times to reach 50% step responses and (2) smaller human-machine closed loop system delays. However, the implications of the relative rapidity of isometric (or stiffer) devices should be interpreted very carefully. Jones and Hunter (1990) also found that as stiffness increases, subjects' accuracy tended to decrease. This means that the shorter 50% response time may not result in better performance. A "fast" system with large overshoot may have a shorter response time, but the final settling time (time to reach and remain within 2% of the final target) could be even longer than a "slower" system. Unfortunately Jones and Hunter did not report on the settling times for each condition tested.
Using a two dimensional positioning task, Mehr and Mehr (1972) did a comparative study between (1) a spring centred joystick in position control mode, (2) an isotonic joystick in rate control mode, (3) a thumb-operated isometric joystick in rate control, (4) a finger operated isometric joystick in rate control mode, and (5) a trackball. It was found that condition (4), which involved an isometric device, showed superior performance (in terms of both completion time and error ) than condition (2) which employed an isotonic device. However, one can not identify the cause of the performance differences since the three factors, i.e., resistance, transfer function and body parts, were all confounded in that study.
Dunbar, Hartzell, Madison, and Remple (1983) presented a comparison study in the context of helicopter control. Conventional helicopters have three separate controllers, namely cyclic, collective, and rudder pedals, controlling pitch/roll, heave, and yaw respectively. Dunbar and colleagues compared a set of conventional separated controls with two integrated controllers, one isotonic and one isometric, in a 3 axis (pitch, yaw, roll) compensatory tracking task. Under all three levels of task difficulty (as defined by bandwidth of the signal being tracked), the RMS tracking errors with the isometric controller were found to be significantly smaller (i.e. better performance) than the RMS errors with the isotonic controller.
Dunbar and colleagues were surprised with the fact that the isotonic controller showed even worse performance than the conventional, separated controllers. The authors speculated on three causes for the results. (1) Display. A 2D, compensatory display was used in the experiment, with pitch error displayed along the y-axis, yaw error displayed along the x-axis, and roll error as angular rotation in the plane of the display. The authors believed that a compensatory display might have suited the isometric controller while a pursuit display might been more suitable for the isotonic controller. (2) Task. The tracked target (signal) had relatively high bandwidth and the isometric controller may have an inherently higher bandwidth than isotonic controllers. (3) Implementation. The gains were not necessarily set at an optimal value for every type of controller.
Ware and Slipp (1991) did an informal comparison study with a
3D navigation task. They used a 6 DOF Spaceballô (isometric)
and a 6 DOF Flying Mouseô (isotonic) to control the velocity
of the user's viewpoint. Subjects were asked to navigate through
a tunnel simulated in a graphical display. They found that on
average the Flying Mouse traversal times were 66% of those obtained
with the Spaceballô. The subjects also did a free scene
exploration task and reported their subjective evaluations. It
was reported that the subjects felt that they were able to control
six DOF simultaneously with the isotonic controller but not with
the Spaceball, with which they could effectively control only
one dimensional translation or rotation at a time. However, the
users also complained of arm fatigue with prolonged use of the
six DOF isotonic device, but not with the isometric controller.
To summarise, the literature on the relative advantages and disadvantages of isometric versus isotonic devices has not been conclusive. Some reports support isometric devices while others support isotonic devices. The definitive answer may depend upon dimensions of the controllers other than resistance and also on the tasks used for the experiment. Bandwidth (response speed) and extent of feedback have been the two major underlying factors that researchers have believed to account for the theoretical differences between isometric versus isotonic devices.
With regards to the response speed, it can easily be concluded that human response with an isometric device is faster than a comparable isotonic device, since no transport of limb or device is needed. However, whether humans can effectively make use of this rapid response, while maintaining acceptable accuracy, is questionable.
With regards to the feedback, in the literature just reviewed,
there appears to be a tacit agreement, either explicitly or implicitly,
that proprioceptive feedback from the control device is a facilitator
of control actions. However, different researchers disagree on
which device actually provides stronger feedback: the isotonic
devices that afford movement cues or the isometric devices that
afford force cues? This question should be addressed in the neuromotor
and psychomotor control literature. More theoretically, whether
feedback is indeed needed for manipulation control is also a relevant
question. This again is in the domain of human motor control.
Part of chapter 3 of this thesis will review some of the literature
in human motor control relevant to input device design and address
these questions.