This is a promised continuation of my earlier post How good software makes us stupid.
I quote from Christof van Nimwegen’s work The paradox of the guided user: assistance can be counter-effective:
A recurring issue in usability guidelines is the importance of minimizing “user memory load”, also referred to as computational offloading (Scaife & Rogers, 1996). This means that the working memory (WM) of a user is relieved so that a maximum of cognitive resources can be devoted to the task. To achieve this, certain information can be externalized to the problem solver and thus be of assistance. Already two decades ago, Norman (1988) proposed the idea that knowledge might be as much in the world as it is in the head. He pointed out that the information embedded in technological artifacts (such as computer interfaces) was as important to task achievement as the knowledge residing in the mind of the user. [...]
Externalization is to make parts of the interface context-sensitive, e.g. by hiding or disabling functions that are not applicable at the moment. By doing so, the user is “taken by the hand” by limiting choices and providing feedback (Van Oostendorp & De Mul, 1999). Examples are wizards, help-options and grayed-out menu-items that do not permit using them, thus offering a context-sensitive interface where only possible actions are displayed. For example, in MS-Word, one cannot select “paste” from the menu when nothing is copied first. “Paste” shown in gray color indicates both the presence and the unavailability of the command. “Graying out” externalizes the applicability of an operator at a certain moment, and makes memorizing this information unnecessary.
In the opposite situation, when no such features are provided, a user has to internalize the information by himself, and store this information in his memory. The above definition of usability implies that systems are tools we use to accomplish a certain goal (text processing, spreadsheets, editing video, looking for information on the web). Rubin (1994) was probably right, though it can be interesting to look at what these presumed friendly systems actually do to humans and their behavior on a deeper level, and whether the highest achievable usability possible in the classical sense of the word is in fact always desirable.
van Nimwegen’s concludes that “externalisation” harms learning. I wrote about this fairly obvious effect in my book Mathematics under the Microscope and apologise for quoting myself:
For some years I had been teaching courses in mathematical logic based on two well-known software packages: Symlog and Tarski’s World [...] Symlog used a DOS command line interface which was extremely weak even by the standards of its time, while Tarski’s World very successfully exploited the graphical user interfaces of Apple and Windows for the visualization of one of the key concepts of logic, a model for a set of formulae [...] . Also, Tarski’s World made very clever use of games to explain another key concept, the validity of a formula in an interpretation [...] However, when it came to a written test, students taught with Symlog made virtually no errors in the composition of logical formulae, while those taught with Tarski’s World very obviously struggled with this basic task. The reason was easy to find: Symlog‘s very unforgiving interface required retyping the whole formula if its syntax had not been recognized, while Tarski’s World‘s user-friendly formula editor automatically inserted matching brackets. Although Tarski’s World‘s students had no difficulty with rather tricky logic problems when they used a computer, their inability to handle formulae without a computer was alarming. Indeed, in mathematics, the ability to reproduce your mental work has to be media-independent. Relieving the students of a repetitive and seemingly mindless task led them to lose a chance to develop an essential skill.[...] we understand and handle much better those processes and actions which we can describe in words. In naive terms, typing a command is like saying a sentence, while clicking a mouse is equivalent to pointing a finger in conversation. The reader would no doubt agree that, when teaching mathematics, we have to incite our students to speak. The tasks of opening and closing matching pairs of brackets, however dull and mundane they may be, activate deeply rooted neural mechanisms for the generation of parsing rules, and are crucial for the interiorization of symbolic mathematical techniques.
In modern GUI’s, users often click around without control, clicking and clicking on and on has become our second nature, perhaps partly due to the World Wide Web. The undo facilities that became almost mandatory in systems are mostly good and handy features, but perhaps they changed our ideas about the consequences of our actions in computer environments.[...] one has to delve into human nature, and be aware of the mechanisms that are natural to us, which include, unfortunately, laziness and shallow behavior. There is nothing we like better than the feeling that things are being done for us, and in many cases, this can be exactly what is desired. However, in some situations one has to be careful to neglect these characteristics. This research concerns exploring the conditions under which motivation and deep contemplation for the task can be provoked to achieve better task performance, fewer errors, and having users that constantly are on top of the task.