Have you ever played one of those candy jar guessing games where the goal is to accurately estimate how many candies are inside to win them all? As a kid, I always tried my hardest to guess correctly because my mom rarely let me have candy. No matter how hard I tried, my guesses were always off, and I never won :(.
Why is making these types of estimations so difficult? When we estimate, we tend to rely on distributed attention, scanning the whole scene rather than focusing on individual pieces. In contrast, counting requires focused attention, which helps us be more precise. Since different attention modes are involved in estimation and counting, factors that influence attention could also affect these processes, potentially leading to biases.

One possible factor? Color. Because color helps guide attention, it could potentially play a role in shifting how we process visual information to make estimations or count. Imagine a candy jar filled with a mix of yummy brightly colored M&Ms like the picture above versus one where the candies are all the same color. Does the variety of color make it harder or easier to estimate and count? Could an organized grouping of color help?
Given the lack on research on how color influences numerical cognition, a recent study published in Psychonomic Bulletin & Review by Qi Li, Guo Ting, Yuichiro Kikuno, Yokosawa Kazuhiko (pictured below) investigated whether color influences attention in ways that affect estimation and counting of numerical quantities.

Experiment 1: Estimation
In the first experiment, their goal was to examine how color variety and spatial arrangement of different colors influences people’s ability to accurately estimate numerical quantities. In each trial, as shown below, participants viewed an array of circles and were asked to estimate the number of circles on the display.

To parse out how color variety influences numerical estimations, on some trials, the circles were all the same color (single color), four different colors (medium variety), or eight different colors (high variety), as shown below.
The researchers also wanted to understand how spatial arrangements of color may influence estimation, so the participants completed two different versions of the task on separate days. On one day, participants viewed clustered arrangements of circles, where all the circles that were the same color were grouped near each other (see the first row below). On the other day, they saw a random arrangement where the colors were randomly distributed without any specific grouping (see the second row below).

In line with previous research, participants tended to underestimate the number of circles when there was a higher number of circles present (right side of the graphs below) and overestimate when there were fewer circles (left side of the graphs below). The authors also found that higher color variety (shown in dark blue on the graphs below), made participants perceive more circles, regardless of whether the colors were clustered or randomly distributed. When the number of circles was small, high color variety exaggerated the existing tendency to overestimate, while for a larger number of circles, high color variety reduced underestimation, improving accuracy.
Participants likely use distributed attention to process global features like color richness, using it as a heuristic to infer a larger number of objects, without being influenced by spatial details.

Experiment 2: Counting
In experiment 2, the authors wanted to investigate how color influences serial counting. The experimental design in experiment 2 was identical to experiment 1, except instead of asking participants to estimate the number of circles present (estimation), participants were asked to count the number of circles present as fast and as accurately as possible (counting).
Contrary to experiment 1, clustered vs. random arrangements of colors did influence counting performance in terms of how fast participants could accurately count. As shown in the graphs below, when the colors were clustered, a higher variety of colors improved counting speed and accuracy, especially when there were larger quantities of circles present. When the colors were randomly arranged, multiple colors actually slowed the speed of counting (accuracy wasn’t affected).
The authors suggest that the clustering of multiple colors can function as visual dividers creating natural segmentations to enhance focused attention. In contrast, when the colors were arranged randomly, they introduced visual noise, which may make focused attention more difficult. This may have been why high color variety in random arrangements slowed counting speed.

While this study might give us some insights on the different biases at play that make the candy jar guessing game so tricky, understanding how color biases numerical cognition has much broader implications. For instance, the next time you’re at the grocery store thinking you’re getting a lot of candy for a good deal, you might want to reconsider whether the colorful candy on display is influencing your judgement. Beyond candy, this research has implications in areas like education, user interface design, or even disaster response, where managing color biases could have a real impact on numerical cognition.
Featured Psychonomic Society article
Li, Q., Ting, G., Kikuno, Y., & Kazuhiko, Y. (2025). The influence of increasing color variety on numerosity estimation and counting. Psychonomic Bulletin & Review, 1-11. https://doi.org/10.3758/s13423-024-02625-x