The Gnome Desktop is a Graphical Environment for the Linux operating system. It is the graphical layer of the operating system that is presented to the user on the screen. The imminent introduction of a new interface in the release of Windows Vista means that Interface Design is currently being heavily promoted and scrutinised. Since 2002, the Gnome Desktop Project have started two sub-projects concerned with usability and interface design. This essay will attempt to outline areas in which developers have utilised cognitive psychology theory, areas in which it could be developed, and areas in which it has been misappropriated.

Information Architects mostly come from backgrounds related to cognitive psychology. Their focus is on understanding the user's interactions with the computer, from mental categorisation to actually learning how to utilise computer programs. Gould and Boies (1983) propose that the user interface should be designed by psychologists, however, due to their being few people with the necessary repertoire of skills, this is unlikely. There is a lack of studies into the decision-making processes in interface design, possibly due to the fact that system development is so confidential. The reason why an “Open Source” Desktop environment was chosen, was the benefit of transparency in development, as all development information is freely available from the GNOME Community.

Interface design is a multi-faceted discipline, with many different cognitive applications. As the desktop interface is a graphical environment, visual perception becomes an important factor in the design. In a graphical environment, it is often visual cues that lead the user through the interface. However, Card et al (1984) suggest there is very little systematic understanding of the interaction between the display and the user's ability to perform cognitive tasks.

Card, Pavel and Farrel (1984) suggest that windowing (the interface employed by Microsoft Windows ©, and also the GNOME Desktop) is more conducive to assisting human performance, as it also accesses to multiple sources of information, whilst allowing these sources to be combined. Windows are also good as reminders, as they function as “breadcrumb trails” in the identification of their content. Recently there has been a new feature added to GNOME, to enable a 3D Desktop. The main feature is that four sides of a 3D cube are used as Desktops, and the user can rotate the cube in order to work on any desktop concurrently. The evolved cognitive abilities of being able to navigate positionally and geographically are now put to use, something that was previously unexploited by a windowed interface.

Gestalt theory espouses that proximity infers commonality. This application of cognitive theory is apparent throughout the interface, with tool bars and menus being placed in horizontal groups. The Gestalt law of similarity is also adhered to, with menus congruently animated, and icon sizes being consistent with the rest of the other icons in their group.

Askwall (1985) conducted a study to identify the ease of reading on a computer screen, in comparison to paper. He found that whilst the reading speed between the two media did not differ significantly, the ability to search through the document did, with many participants only being able to search through half as much material on screen as they did on paper. However, there have been many advances in screen technology since the study was conducted in 1985, with font rendering improving and the introduction of more colourful and dynamic displays.

Raskin (2000) argues that the purpose of an interface is to enable an automation of a task. The theoretical concept of neural pathways are echoed in the idea of “global functions.” Functions that are used in one interface should be consistent across the rest in order to not only save on learning time, but reduce conflict. If a particular function has one outcome in one program, and a different outcome in another, it will be easier to forget, or at worst misappropriate. Withrow (2003) postulates that this concept can be defined as transference; those functions that are congruent with past experiences. Positive transference when the function behaves as expected, and negative transference when there is a lack of congruence.

The execution of computer-based procedures changes systematically as learning progresses. A QWERTY keyboard may take many weeks to learn, yet someone with experience at using on may find it an intuitive interface for typing. Shiffrin and Dumais (1981) argue that most skilled activity is underlyed by automatic processing, highlighting that automation is no necessarily the skill itself, but a component of the skill, as skilled performance is more than the sum of its automated components (Logan, 1985).

Schneider and Fisk (1980) note that the automation of action is globally acquired, in that one learns how to catch an object, rather than learns how to catch a specific ball thrown at a certain trajectory. This has particular relevance to the GNOME Desktop environment, as many people using it have come from a Microsoft Environment. The main global functions are similar to those used in Microsoft Windows, however, there are also many differences. However, it has been noted that most functions one learns on a Microsoft Machine can be transferred across to the GNOME Linux Desktop.

The transition from declarative to procedural knowledge, or explicit to implicit knowledge is of major concern to user interface designers. Human Computer interaction is never going to be perfect, due to the “human factor.” However, consistency, such as knowing the exact behaviour of the delete key in any application can make such activities appear automated and skilful. The designers' task is therefore to create an interface that is as automatic as possible, without becoming rigid and insensitive to unusual circumstance.

A good example of correctly applied cognitive theory is that of a deletion message (Gardiner and Christie, 1987). Currently, when a user is about to delete an item, a message pops up saying “Are you sure?” advising the user to click yes or no. Due to the fact that the user has automated the process of clicking “Yes” to get through dialogue boxes, there is a high likelihood that the user will click “Yes” without reading the message. A better interface from a cognitive perspective would be to ask the user to press <x> to complete the command (where <x> is a random letter).

Cognitive studies into the acquisition, storage and retrieval of memory are plentiful, as memory is concept integral to cognitive psychology. There have been many applications of cognitive studies on memory in interface design. Some designers took the work done by Miller (1956) on the size of short term memory, and applied it, incorrectly, to the number of options that should be in a menu. However, because the menu is always on display, this research has more to do with the number of pages/links one can click through before forgetting where one started.

The way in which data entered into a computer is handled is not conducive to a good interface, according to Raskin, (2000). He argues that information entered into a computer should be treated as sacred, and should be fully recoverable, even after deletion. Whilst this has obvious implications from a physical storage perspective, it poses an interesting concept. Computers are one of few things that one can put something into, and it be lost completely.

Current computer design means that the user has to learn the interface. Whilst this may have been acceptable for the pioneering work done in information technology, Mahdi Abdulrazak (2004), a member of the Iraqi Linux Users Group, writes that perhaps the interface should learn the user. An example of this concept is a program written for disabled co
mmunication company, ACE. Once the user has selected a letter on the touch screen, the letters more commonly used next then appear bigger. This has major advantages from people with deficient motor reactions, as the interface also learns words that the user uses more often, making these sequences more prominent in selection.

The actual physical input of data into a computer is something that has been studied extensively. Card, Burr and English (1978) demonstrated the superiority of the mouse over the joystick, in the context of deriving models for the prediction of human performance, whilst Goodwin (1975) had demonstrated the efficiency of the light pen and gun over a badly designed keyboard. Haller, Mutscler and Voss (1984) concluded that it was the compatibility of the light pen, giving feedback is visual tactual and proprioceptive modes, that inferred its superiority. Finney (1997) conducted an experiment on the effect of Delayed Auditory Feedback on rhythm-tapping, and found that the delay causing maximum impairment, was not fixed, but was rate-dependent. This has consequences for Human Computer Interaction, indicating that consistency is more important than immediacy for data feedback.

There is a wide variety of cognitive theory being applied to information architecture. Rather than just creating an interface which interfaces the code with a graphical display, designers are being more directed to interfacing with a user (Nilsen, 1999), with many project now running with design consultancy from cognitive psychologists in order to understand the human computer interaction. Whilst there is still a lack of necessary skills in this regard, it is starting to be developed further, with the introduction of the 3D desktop, and new studies into input methods and user feedback. One of the biggest advantages of the GNOME Desktop is that of user feedback, in that users are directly contributing to the development of the interface. Whilst in isolation, this may prevent the introduction of true innovation, it serves as an empirical feedback mechanism for developers to further tailor their interfaces to the users, as recommended by cognitive psychologists.