Procedure

Participants were instructed to study the map stimulus – as long as they wanted - to be able to remember the main structuring elements (rivers, roads, water bodies, etc.). Once they think they studied the map long enough, they pressed a certain key as instructed beforehand and exited the first part of the assignment. Next, they had to draw this map from memory by using MS Paint as a sketching tool. This tool was selected, because both experts and novices would not need any prior training. After the execution of task, in order words, drawing the sketch map, participants used a special key to terminate the task. There were not any time limitation either for studying or drawing procedures. While participants studied and drew the map, their eye movements were recorded.

Apparatus and Recording

The experiment was conducted in the Eye Tracking Laboratory of the Marketing Department of Ghent University. We recorded participants’ eye movements using a SMI RED 250 eye tracker mounted to the stimulus monitor. The stimulus was shown on a 22” colour monitor with 1680 x 1050 spatial resolution.

Methodology

Sketch maps

To be able to analyse sketch maps, we needed to quantify the information presented with them. Hence, we first determined the structuring map elements on the original map/stimulus, then counted and classified them into four main categories; hydrology, land-cover, settlements, roads. The original map was consisted of four hydrographic features, four land-cover features, eight residential areas/settlements, ten roads (in total 26 map elements).
The sketch maps were analysed based on the literature on cognitive process (see literature review) and the literature on sketch map evaluation for cartographic usability. In this context, we identified two main criteria; drawing order and the score on drawn elements. Drawing order information derived from the registered eye tracking video and was thus, assessed individually with an attempt to identify the hierarchical structure throughout the cognitive process of spatial information. The scoring of the map elements considered position (presence and accuracy), size, shape and colour of the drawn elements and corresponds to the qualitative characteristics of the sketch maps. It seeks an answer for how well the sketch map is executed (complete and accurate), and accordingly how well the cognitive map is constructed. This scoring method will be explained with more detail in the results.

Eye tracking

We employed eye tracking to extract the drawing order information, as well as we evaluated relevant eye tracking statistics, since we are aware of the significance of these in understanding/revealing cognitive processes of participants. Eye tracking statistics such as number of fixation per second, the average duration of fixation, and trial durations for study and drawing time, helped us to evaluate the performances of participants.
Similar to \citet{v2015} , number of fixations per second were taken into account instead of fixation count, which corresponds to the fixations for the entire trial. The reason is that the fixation count relies on the length of the trial and since every participant completes the task at different time span, we needed an objective measure to compare the performances of our target groups. The number of fixation per second for a participant is calculated by dividing fixation count by trial duration. There is a strong relationship between the number of fixations per second and another often used metric; average fixation duration. The longer the fixation durations, the fewer the fixation per second. Fixation duration is also linked to the cognitive processes of the visual stimulus. Longer the fixations may indicate that reading the map gets harder which causes a rise in cognitive load ( \citealp*{2007} ; Ooms et al., 2014) or it may be interpreted that the user finds the map or a certain part of it interesting (\citealp*{k.2012} ). People also concentrate their fixations to most informative parts of the visual stimulus (\citealp*{f.2004}).
Lastly, trial durations of participants were taken into account. Although there was no time limitation for both study and drawing parts of the memory task, trial times give insight about motivation and top-down attention. Inherently, longer trial durations for studying the map indicates higher level of interest.

Results

Sketch map evaluation

Drawing order

\citet{k.1960} described the sequential process of sketch map creation as the order that each element in a sketch map is drawn. Since sequence includes space and time, the sketch maps reveal the underlying cognitive process of ordered retrieval of the movements within time and space, which are resulted as our cognitive maps. Having had this unique characteristic, sketch maps can be treated as three-dimensional representations through space, time and sequence ( Huynh and Doherty, 2007). The hierarchical order of nodes and paths drawn on the sketch maps represents the hierarchical order of information (primary-level, secondary-level, and so on) presented on the maps. The earlier the element is recalled, the more important it is to a person. Lower hierarchical levels correspond decreasing amount of spatial information, decreasing frequency of use and greater difficulty of remembering. (\citealp*{Golledge_2010}). Hence, drawing order can give insights in how these elements were stored in the users’ memory. Elements that are first drawn are more accessible in the LTM; they can be retrieved more easily (\citealp{v2015}). We extracted the drawing order information by analyzing scan paths of participants via eye tracking. Figure 3 depicts examples of sketch maps drawn by experts and novices as a result of memory task.
For the assessment of drawing order, we implemented the scoring system used by  \citet{v2015}. As we classified the main structuring elements contained by the map into four categories, we assigned a score to each category depending on the order in which the first element of that category was drawn. The scoring was 100, 50, 25, 5 for the first, second, third, and fourth, respectively. If a certain element did not exist on the sketch map, we assigned zero as a score.