Aversive bimodal associations impact visual and olfactory memory performance in Drosophila

Devasena Thiagarajan, Franziska Eberl, Daniel Veit, Bill S. Hansson, Markus Knaden, Silke Sachse

Preprint posted on 23 July 2022

Article now published in iScience at

If seeing is believing, then seeing and smelling is learning: memory in Drosophila with bimodal cues.

Selected by T. W. Schwanitz

Updated 21 February 2023 with a postLight by Timothy W. Schwanitz

The preprint discussed here was recently published in iScience. It was a good article already in its preprint form, and so no major changes were necessary for publication. The new graphical abstract is one of the nicest additions to the publication, especially since it manages to boil down the complexity of the results into a relatively simple graphic. Some of the nuance and subtlety is preserved even in the simplified graphic, e.g., the preference wheels in the graphical abstract indicate partial preference in a way that is roughly proportional to the actual data. One cool result from the study that sadly did not make it into the abstract: the differences in the interactions between two senses in long-term bimodal memory vs short-term bimodal memory. Readers will have to dig deeper than the graphical abstract to learn about those interesting findings.

No new figures were added, though Figure 1 got split into two, and two panels were added to now Figure 2 to further clarify training paradigms. Some of the biggest changes to the paper were in the discussion, which now does an even better job of situating the broader impact of the findings (I smiled to see that the phrase “more the merrier” from my preLight made it into the discussion). The discussion of the final published paper is clearer and helps the reader better understand what is surprising in this study. The addition of a study limitations section also assists the reader in interpreting the results.


Think of a memorable place you have lived. Do you recall sights, sounds, smells, or maybe even textures and flavors? Memory often involves the complex integration of several senses. Many studies of the fruit fly Drosophila melanogaster have investigated their impressive olfactory learning abilities, or they have studied the fly’s visual system via color learning; however, not as many studies have looked at combinations of sensory inputs. When studied in isolation, fruit flies have often fared poorly in learning trials using color alone as a training cue (relative to their olfactory abilities, that is).

Thiagarajan and colleagues therefore investigated two sensory modalities, vision and olfaction, and their joint influence on learning. The researchers used a T-maze with four different stimuli: two food odors, acetoin and ethyl butyrate, as well as two different wavelengths of light, green and blue. The authors optimized the odor concentrations such that neither odor was inherently more attractive to the flies, and they did the same with light intensities. Flies were trained using the aversive stimulus of an electric shock in a training tube separate from the maze. The scientists used several different learning paradigms, all assessing how effectively the flies can recall the aversive association with either an odor or light stimulus—as measured by how many flies avoid the part of the T-maze with the aversive stimulus (see Fig. 1B of the preprint).


Fig. 1B. A schematic of the training tube that was used prior to the T-maze. Flies were exposed to an odor, a color of light, or a combination of both, while also being given an electric shock for one minute per round of training.

Key findings:

Flies are good at learning olfactory associations—but they are not as good at learning visual associations. The authors found no difference in avoidance between flies conditioned on an aversive olfactory stimulus in just one trial or in four trials. By contrast, a single learning trial was not sufficient to get a robust avoidance of the color of light associated with an aversive stimulus. After four learning trials, however, the flies did avoid the color of light associated with an aversive stimulus.

Having come to these conclusions, the researchers then did a series of tests where both an olfactory and a visual cue were used during the learning trials.

Olfactory cues improve visual memory; however, visual cues do not improve olfactory memory. If flies were trained on both an olfactory and a visual cue, but were then tested only on the visual cue, they performed better than if they were trained exclusively on visual cues; hence, olfactory cues augment visual memory. Nevertheless, flies trained on both cues but only tested on olfactory cues did just as well as flies trained only on olfactory cues. It appears that olfactory memories are already so strong that they cannot be enhanced by visual cues.

After these findings, the authors created a more complex learning environment where the flies were trained bimodally on all possible color and odor combinations. For example, acetoin and blue were presented together with an electric shock as the conditioned aversive stimulus, while ethyl butyrate and green were presented without a shock. The flies were then tested on only the visual or the olfactory cues.

Visual memory is improved by the addition of olfactory learning cues. In congruence with the previous experiments, flies that were trained with odors as well as colors learned to associate colors alone with a negative stimulus, while flies that only had colors did not form a significant visual memory. These results were confirmed by also using transgenic flies (i.e. olfactory mutants) that could not detect the odors used in the trial. Transgenic flies performed like flies that were trained only on colors, underscoring that visual memory formation is not very strong unless augmented by olfactory inputs.

Olfactory memory can be diminished by the addition of visual learning cues. Strikingly, though, flies that were trained on both colors and odors, and then tested only on odors, did not perform as well as flies that were only trained on odors. Adding a weak visual memory stimulus can actually reduce the strength of the olfactory memory.

The above results were all looking at short-term memory effects. As a final flourish to their study, Thiagarajan and colleagues added long-term versus short-term memory.

Long-term olfactory memory was improved by the presence of visual cues. In contrast to their short-term memory focused findings, adding a visual cue to olfactory learning improved long-term memory: more flies avoided the aversive stimulus. This result is especially interesting given that the flies had no detectable visual long-term memory, i.e., just using a visual stimulus after bimodal training did not result in significant levels of avoidance.

Why I think this study is important:

Memory is a complex process that involves time and many senses, so integrating multiple senses into experimental designs is an important step toward gaining a better understanding of how memory actually functions. The results in this preprint are especially fascinating because they show how memory is not a simple case of “the more the merrier.” Sometimes additional senses improve memory, as when the authors augmented visual cues with olfactory cues and then tested the flies in the short-term, or when the authors added visual cues to olfactory cues and tested the flies in the long-term. Other experiments in this study, however, showed that additional sensory cues weaken memory associations, as when the researchers added visual cues to olfactory cues and tested the flies in the short-term.

These complex results underscore the importance of testing assumptions like the idea that more sensory modalities always translate to better memory skills, or that findings at one temporal scale will hold true for all temporal scales.

Tags: behavior, color, drosophila, memory, olfaction, t-maze, vision

Posted on: 9 August 2022 , updated on: 21 February 2023


Read preprint (No Ratings Yet)

Author's response

Devasena Thiagarajan and Silke Sachse shared

1. You used an aversive learning set up, where the flies were trained to associate a stimulus with a negative memory. Do you think your results could have been different with a positive stimulus, e.g., a sugar reward?

Several studies in honeybees (proboscis extension response) and in ants (nest identification) have used multimodal stimuli associated with sucrose rewards and have also observed similar effects, i.e., boosting of learning performances, especially when working with weakly learned unimodal cues. Colors and odors have also been used along with sucrose rewards in Drosophila, where enhancement (but not attenuation) was observed (Okray et al., 2022 – BioRxiv preprint).

2. Do you have any plans to add a third sensory modality, e.g., sound/vibrations, or perhaps texture? Given the challenge of adding yet another modality, and the additional number of possible combinations this would create, do you think a T-maze would be an optimal set up for such a study?

The copper grid that lines the insides of the tubes provides a mechanical cue which also serves as a context that we make consistent between the training and the testing phases. However, by decoupling it, we can make it an exclusive mechanical cue that can alter learning performances. The work done on context dependent LTM emphasizes this effect clearly (Zhao et al., 2019). Therefore, we believe that the T-maze is simple and robust enough for studying vision/olfaction and mechanical stimulation, but agreeably, experimental complexity will be increased when you add more stimuli to the mix in a T-maze.

3. You used a continuous and even color as your visual cue. Do you think that the addition of shapes could improve visual memory performance?

We wondered along the same lines, especially to see if one form of visual learning (patterns, shapes etc.) can be used to enhance another form of visual learning. Although we have not conducted those experiments, we can derive some information from a study done on mantis shrimps. It shows that there is inherent preference to form better learned associations with shapes than with colors, indicating that when same reinforcements are made with redundant cues, there is a preference over the better learned cue than a synergistic outcome with both (Patel RN. et al., 2021).

Have your say

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Register here