In the paper "Display-based action at the user interface" (Payne, 1991), Stephen Payne proposes a model for planning in the interaction between a user and a computer. The model is based on the assumption that the user does very limited pre-planning and he decides his future actions based on the information displayed on the screen. This means that the information flow in the interaction is not only from the user to the device (the operators) but also from the display to the user (the feedback). The model conflicts with other models that assume that the typical user plans in detail before he actions. It is close to the cognitive model called Opportunistic Planning, proposed in Hayes Roth and Hayes Roth, 1979, and presented in the previous section.
To support his model, Payne presents two empirical studies. The purpose of these experiments is to show that the assumptions of the model are correct, that is, the user does limited pre-planning. To show this, Payne goes even further by stating that the user cannot do much planning because they do not know in details the effects of the operators used in the interaction with the computer. We can summarize the inference used by Payne in the following statement.
Experiments show that even experienced users are uncertain about the effect of the operators. That implies that they cannot plan in advance the actions required to accomplish a task. So, to achieve the desired goals, planning and action are interleaved and to a large extent decisions are made based on what is displayed on the screen.
The rest of this section presents the studies from Payne's paper.
Description: The users were asked to answer a questionnaire. The questions asked the users to determine the position of the cursor in an edited text after a number of editing commands. For example:
Question: Where the cursor will finish up if a back-a-word command is issued in the following situation (the current position of the cursor is marked with bold):
Results: Only a small number of users know the word-based commands and many do not know the exact behavior of the word processor when the text contains special characters (e.g. dash).
Interpretation:
Purpose: Test the knowledge of menu-driven user interface for a word-processor.
Description: The subjects were asked to answer several questions about where is the cursor and the state of the interface during a find-a-word operation.
Results: Few subjects could predict the correct behavior. They do not know all the details of the dialog for searching. For example, the used did not know what is the label on the buttons used in the search procedure.
Interpretation: The users recall only patchy the dialog window for search. They are uncertain about: where the cursor is, what part of the text is selected, what happens if a new word is typed in.
The two studies show that:
The conclusions of the two studies give support for the model proposed by Payne, the Display-Based Action Model. The users depend heavily on the information displayed. They do limited pre-planning before starting a task and many decisions are taken between the actions and based on the state of the screen.