Identifying maximum/minimum values in different charts is a common goal when users are interacting with visualizations. Moreover, this can be a challenging task when the user is interacting with space-filling visualizations. Space-filling methods consider a 2D area to represent values, however, low values that might be interesting for decision makers are commonly obfuscated by such approach. In this context, we conducted an eye tracking study with 12 target users to analyze how scale inversion (area proportional to 1/value) can support the identification of elements with small values represented in treemaps. Results show that when the inverted scale supports the task, users are able to identify elements individually and in a less ambiguous way, but took more time to interpret the inverted scale treemap. With the presented results, we expect that Human-Computer Interaction practitioners working on space-filling visualizations to include interactive scales to invert the visualizations scale when the task at hand involves evaluating minimum and/or maximum values.