The control of self-motion is a fundamental ability. To understand the control of self-motion, and indeed action in general, it is useful to distinguish between variables which are manipulated by the "controller" (control variables) and variables which the controlled is a attempting to make behave as desired (controlled variables).

Control variables may be thought of system input, while controlled variables are the system output. Control variables are the means by which the controller gains mastery over the controlled variables.

Successful mastery of a controlled variable implies at least one of the following: (i) the ability to determine the consequences of actions already taken (feedback information); (ii) the ability to anticipate what is going to happen.

Feedback control

The use of feedback information to control the behaviour of a system in the presence of disturbances is commonplace. This type of control is sometimes called closed-loop.

Automatic control systems can be constructed which do not require human intervention. A simple closed-loop negative feedback control system consists of the following components:

1. System being controlled (plant)
2. Sensor to measure an output variable (controlled variable)
3. Differencer or comparator
4. Amplifier
5. Actuator

For example, a system for the automatic control of the temperature (controlled variable) of a room (1) consists of a method for heating or cooling the room such as an air conditioner (5), a sensor for measuring the current temperature (2), and thermostat for comparing this to the desired temperature (3), and an appropriate switch to convey the result of this comparison (4). When the temperature of the room increases above a threshold value the thermostat registers the positive difference and causes a response of opposite sign, that is, activates cooling.

Automatic control of a plant is often too difficult, or too expensive, and a person may be involved in the control system. A person can play the role of one or more of components 2, 3, 4, or 5.

The greatest difficulty for automatic control is often the problem of obtaining the sensory information required, and comparing this with the desired outcome. People have very sophisticated perceptual systems for obtaining such information which are far beyond even the most advanced technology, and this is often the role in which people are utilised in technological systems.

However people do have limitations. These include:

1. Restricted ranges of sensitivity
2. Limitations of attention (sustained and selective)
3. Perceptual illusions
4. Internal time delays
5. Stimulus-Response compatibilities

Control of heading

The control of the direction of travel (heading) is a task which is well suited to closed-loop negative feedback control. To maintain travel in the desired direction in the face of unpredictable perturbations the controller must perceive the current direction of travel (the controlled variable), constantly compare this with the desired direction, and make appropriate (negative) adjustments to the steering mechanism of the plant (the control variable) in response to discrepancies. (See Wallis et al 1997 for evidence regarding the closed-loop nature of steering).

People are very good at perceiving heading, even when there are no obvious cues from a Head up display, car hood, or even road edges or lane markings.

A general source of heading information is available which might account for this ability ­ the focus of expansion of the optic flow field.

When an observer moves in a stationary environment, the light reflected to the moving eye from elements of the environment undergoes a lawful transformation over time called the optic flow. Optic flow may be represented by an instantaneous two dimensional velocity field (see example below) where each vector corresponds to the optical motion (magnitude and direction) of each environmental element.

Example of optic flow in a virtual "star field". Longer lines indicate higher velocity

The optic flow provides information about the layout of surfaces and objects (previously referred to as motion perspective), and information about the motion of the observer. For example, the focus of expansion of the optic flow (where there is zero optical motion) specifies the current direction of travel of the observer (heading).

Access to the information in the optic flow is somewhat complex. Unless the observers eyes are undergoing translation only (no rotation due to head movement or eye movement) the visual flow at the retina (the retinal flow) is not identical to the optic flow. In the general case, the retinal flow is a consequence of both observer motion and eye motion. Humans are able to effectively disambiguate the sources of retinal flow to obtain heading information, although whether the retinal flow alone is sufficient to obtain this information remains a subject of debate and a recent review (Lappe et al., 1999) suggests that extraretinal information is combined with retinal flow to determine heading in at least some situations. (see Bankslab web page).

Perceptual illusions in the control of self-motion

There are multiple sources of information which may be used in the control of self-motion. Vestibular, proprioceptive, and auditory information is available in addition to visual information. In general, stance and locomotion are regulated by visual and proprioceptive information, while vestibular information contributes to recovery from high frequency perturbations.

Under normal environmental circumstances information from different sources is combined to provide a veridical perception of the information required for the control of self-motion. Under abnormal circumstances perceptual illusions may be encountered with potentially serious consequences.

One abnormal situation in which a dangerous perceptual illusion occurs is the situation of a pilot taking off in conditions in which visual cues are reduced. The classic case is a take off at night into a dark sky with no visible horizon (unlit terrain). This is a potential problem at rural airstrips and during catapult launch from aircraft carriers.

The source of this "somatogravic" illusion is information provided by the otoliths in the ears, part of the vestibular perceptual system. Under normal ecological conditions the otoliths are deformed by gravity in response to changes in the orientation of the head and thus provide information regarding the pitch of the head.

However, large horizontal accelerations like those which occur during aircraft takeoff also causes deformation of the otoliths in the same direction as that caused by upward pitch of the head. The consequence during takeoff is that the information provided by the otoliths specifies an upward pitch which exceeds the pilots actual pitch. In the absence of visual cues, the pilot who is unaware of the illusion corrects by lowering the pitch of the aircraft. This in turn increases the linear acceleration of the aircraft which exaggerates the illusion further, causing further corrections. If the illusion is not recognised the pilot flies the aircraft into the ground at full power. (See Watson, 1992b for a detailed description of this illusion, and Watson 1992a; & 1992c for other examples, also http://john.berkeley.edu/Projects/SGIllusion.html).

Anticipatory control

The second type of information which can be used to allow the control of self-motion (and action in general) is information which allows anticipation of future events. Anticipatory information may be:

1. Anticipation of a disturbance to an on-going behaviour (feedforward control)
2. Anticipation of where or when something will happen (i.e., prediction).

A critical part of control during self-motion (and of other tasks such as those involving the interception of moving objects is accurate timing. Anticipatory information is necessary for the controller to know when to start, for example, braking, or turning. In each case the control of action requires about time remaining to to arrival (either the arrival of a moving object, or the arrival of a moving observer).

The time remaining until arrival could be calculated by the observer first obtaining information about the distance of the object from the observer, and the relative velocity of approach and then dividing distance by velocity.

However, the time remaining until arrival is available from the optic flow without any such calculations being required. The image of an approaching object increases in a systematic way, such that the inverse of the relative rate of dilation of the image of an object approaching with constant velocity provides accurate information about the arrival of the object at the plane of the eyes. It has been frequently proposed (eg., Lee, 1976) that this source of information, termed tau, provides the informational basis for the control of interceptive actions.

Increasing image size of an approaching object

A recent review (Tresilian, 1999) provides a detailed discussion of the limitations of tau as a source of information about time remaining to arrival. These include the neglect of acceleration and that tau only provides accurate information about time to arrival at the eye. While tau is involved in the perception of time remaining until arrival in many circumstances, like the cues for layout, tau is normally combined with other sources of information in the perception of time remaining until arrival.

Required Reading

Schiff & Arnone, 1990 pp 1-35,

Lappe et al., 1999

Tresilian, 1999