Definitions and Distinctions

Continuous vs. Discrete: An input interface which allows and senses a graded, infinitesimally changeable range of the some motor variable – e.g. the position of a computer mouse, the force on an isometric joystick, the loudness of a vocal sound, or the breath pressure at a mouth tube – is called a “continuous” input interface. Two reasonably synonymous terms are “proportional” and “analog”. If on the other hand, an input interface offers a countable set of distinct acts that it will respond to; this is known as a “discrete” interface. Switch arrays, on-screen keyboards, lapboard buttons, toggle switches, and joysticks that only recognize eight positions are all examples of “discrete” interfaces. Interfaces in which each input target evokes a particular output action of the AT system are said to be controlled via “direct selection”.

The distinction between discrete and continuous can get muddy if the action the user performs is continuous but the sensing scheme is discrete. The user of a mouth stick moves it over a continuous volume of space in front of her/his face, but the interface is still discrete if it only recognizes and distinguishes among the activation of, say, sixteen different keys on a four-by-four grid.

Coding: If an AT user is capable of using a discrete interface – consisting, for example, of eight big switches mounted for accessibility to various parts of her/his body – this may not be enough for all the communication or control inputs the user needs in order to make use of some AT system. An English-speaking individual with complete command of conventional spelling needs to access at least the 26 letter of the alphabet plus various punctuation and control inputs. Think of the number of items – roughly seventy – on a standard QWERTY keyboard, for example. One way to make this possible with, e.g., eight switches would be to apply a code. Two strikes of eight switches provide sixty four unique permutations. Three strikes offer five hundred twelve permutations.

The issues in design or selection of coded interface may be complex and certainly include the cognitive demand imposed by learning a code, and/or the technological challenge of displaying the code in an intuitive way that effectively reminds and prompts the user. Despite these real issues of strategy, a coded input interface may allow many unambiguous selections with a minimum of distinct acts. A traditional example is the Morse code which can be implemented with only two switches, one signifying dot, and one for dash.

Scanning: This is a somewhat more commonplace approach and one which requires an absolute minimum of motor control. A scanning interface is one in which a menu display of possible selections (actions for controlling home lighting and appliances in an environmental control unit, for example) is continuously sequenced or scrolled or highlighted in view of the user. When a desired item is indicated, the user activates a single switch as an “enter” or “activate” or “do it” command. Most commonly, scanned arrays are arranged as grids in which one row of items is highlighted at a time. This row is selected with one switch closure; subsequently a second closure is needed to select a single item as that row is scanned.

The obvious advantage of scanning interfaces is that they can provide unlimited access to language and/or control over many AT outputs with only a single input act. The disadvantage, when there are other choices, is that this interface strategy may be very slow. It is almost never useful for input to AT – steering a power wheelchair, for example – that requires fast and accurate timing.