Serial Behaviour: What can fruit flies teach us about karate?

  1. Helen H Yang
  2. Thomas R Clandinin  Is a corresponding author
  1. Stanford University, United States

When performing kata, a skilled martial artist flows through a sequence of individual movements—attacks and blocks, turns and jumps—to defeat a series of imaginary opponents. Of course, such action sequences are not restricted to either karate or humans. Instead, serial behaviours are widespread throughout the animal kingdom, with examples including navigation, courtship and communication. However, it is not clear how a nervous system is able to generate sequences of actions.

Now in eLife, Andrew Seeds and co-workers—who are all at the Janelia Farm Research Campus—have taken a novel approach to investigate serial behaviours by using the powerful genetic tools that are available in the fruit fly Drosophila (Seeds et al., 2014). Two different procedures, or algorithms, have previously been proposed for the production of sequential actions (Houghton and Hartley, 1995). ‘Response chaining’—in which each action triggers the next like a falling row of dominos (Adams, 1984)—is thought to underlie birdsong (Long et al., 2010). Alternatively, all of the actions in a sequence could be prepared in parallel, with an inbuilt order of priority and a winner-take-all competition leading to one action being performed at a time and in the right order. Typing and speaking, for example, are human behaviours that are thought to be implemented through this ‘competitive queuing’ model (Lashley, 1951). However, it has been difficult to obtain causal evidence that would favour one of these algorithms over the other. Now Seeds et al. demonstrate that competitive queuing produces the sequential grooming behaviour observed in fruit flies.

Grooming is a patterned, or stereotyped, sequence of cleaning movements performed by most animals with limbs. Indeed, fruit flies use their front or hind legs to clean specific body parts when they are dirty (Szebenyi, 1969). Seeds et al. completely covered flies with dust and observed that each fly preferentially cleaned its eyes first, followed by its antennae, its abdomen, its wings, and finally its thorax. These observations reveal two points: first, grooming in dirty fruit flies is a serial behaviour; and second, it follows a stereotyped order.

So, how does the fruit fly's nervous system produce this serial behaviour? To address this question, Seeds et al. first identified populations of neurons that, when activated, were able to trigger grooming of each body part in the absence of dust. This revealed that at the neuronal level as well as at the behavioural level, grooming is comprised of individual, separable units or modules. Then, rather than asking which neurons are involved in the cleaning of each body part, Seeds et al. took a novel approach towards understanding the processes, or computations, that underpin grooming and instead asked: what algorithm does the brain use to generate this sequence?

If response chaining were used to control this behaviour, grooming of one body part would trigger the grooming of the next one in the sequence. However, this was not the case. Instead, Seeds et al. show that it is the stimulus—the presence of dust—that drives cleaning of each body part. The cleaning of different body parts, however, is not only affected by dust, since the grooming of one body part blocks the grooming of body parts that follow later in the cleaning sequence. For example, when Seeds et al. covered a fly with dust and activated its abdomen-grooming module at the same time, the fly cleaned its eyes, antennae and abdomen, but never cleaned its wings and thorax. When eye cleaning was activated, the flies continuously groomed their eyes even though there was dust covering the rest of their bodies too. This demonstrates that the order in which body parts are groomed reflects an underlying set of priorities. That is, higher priority behaviours (such as eye cleaning) suppress lower priority behaviours (such as abdomen cleaning), but not vice versa. Thus, these modules interact in a ‘suppression hierarchy’.

The results indicate competitive queuing, and Seeds et al. used a simulation to demonstrate that such a model is sufficient to reproduce sequential grooming behaviour (Figure 1). In a dirty fly, all of the cleaning modules are activated in parallel, but they are prioritized by a suppression hierarchy; a winner-take-all competition then selects the highest priority module, and this module is performed. Once dust is cleaned from that body part, the stimulation of the corresponding module is also removed, and the next most important module is performed.

Competitive queuing produces sequential grooming behaviour.

Seeds et al. suggest that grooming behaviour in fruit flies uses a three stage algorithm to select which cleaning module to perform: sensory input (top), hierarchical suppression (middle), and a winner-take-all competition (bottom). The dust (yellow dots) activates each cleaning module. When the fly is completely covered with dust, all the modules receive sensory input and are activated in parallel. The ‘hierarchical suppression’ stage determines the degree of activation for each module (represented by the area of the coloured region in each circle) in order to implement the order of priority. The most active module is selected and the corresponding body part is cleaned, with all other cleaning modules being suppressed. As dust is removed from a body part, the sensory input to that module is reduced (represented by the blunt arrow). Eventually, this module is no longer the most active module and a new module is selected. Multiple iterations of this process produce the sequential grooming behaviour observed in fruit flies.

Seeds et al. suggest two possible ways of implementing hierarchical suppression. A module could be high priority either because it is more sensitive to dust or because it inhibits the downstream, ‘low priority’, modules. However, further study is needed to determine which is the case for grooming behaviour in fruit flies. Furthermore, the cells and circuits that implement grooming behaviour also remain to be identified.

This study provides causal evidence that modules being activated in parallel combined with hierarchical suppression can produce serial behaviour. More broadly, it also illustrates the usefulness of studying model organisms, as it is remarkable that the same algorithm that is thought to underpin complex, learned behaviours unique to humans (such as typing and speech) also produces a widespread behaviour that is innate in fruit flies (i.e. grooming). Perhaps this algorithm is evolutionarily ancient and has been elaborated in humans, in which case, understanding computation in fruit flies can indeed teach us about karate.


    1. Houghton G
    2. Hartley T
    Parallel models of serial behavior: Lashley revisited
    Psyche 2:1–25.
  1. Book
    1. Lashley KS
    The problem of serial order in behavior
    In: Jeffress LloydA, editors. Cerebral mechanisms in behavior, the Hixon Symposium. Oxford: Wiley. pp. 112–146.

Article and author information

Author details

  1. Helen H Yang

    Department of Neurobiology, Stanford University, Stanford, United States
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5140-9664
  2. Thomas R Clandinin

    Department of Neurobiology, Stanford University, Stanford, United States
    For correspondence
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6277-6849

Publication history

  1. Version of Record published: August 19, 2014 (version 1)


© 2014, Yang and Clandinin

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.


  • 1,736
  • 57
  • 1

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Helen H Yang
  2. Thomas R Clandinin
Serial Behaviour: What can fruit flies teach us about karate?
eLife 3:e04040.

Further reading

    1. Neuroscience
    Zilu Liang, Simeng Wu ... Chao Liu
    Research Article

    People form impressions about others during daily social encounters and infer personality traits from others' behaviors. Such trait inference is thought to rely on two universal dimensions: competence and warmth. These two dimensions can be used to construct a ‘social cognitive map’ organizing massive information obtained from social encounters efficiently. Originating from spatial cognition, the neural codes supporting the representation and navigation of spatial cognitive maps have been widely studied. Recent studies suggest similar neural mechanism subserves the map-like architecture in social cognition as well. Here we investigated how spatial codes operate beyond the physical environment and support the representation and navigation of social cognitive map. We designed a social value space defined by two dimensions of competence and warmth. Behaviorally, participants were able to navigate to a learned location from random starting locations in this abstract social space. At the neural level, we identified the representation of distance in the precuneus, fusiform gyrus, and middle occipital gyrus. We also found partial evidence of grid-like representation patterns in the medial prefrontal cortex and entorhinal cortex. Moreover, the intensity of grid-like response scaled with the performance of navigating in social space and social avoidance trait scores. Our findings suggest a neurocognitive mechanism by which social information can be organized into a structured representation, namely cognitive map and its relevance to social well-being.

    1. Neuroscience
    Alina Tetereva, Narun Pat
    Research Article

    One well-known biomarker candidate that supposedly helps capture fluid cognition is Brain Age, or a predicted value based on machine-learning models built to predict chronological age from brain MRI. To formally evaluate the utility of Brain Age for capturing fluid cognition, we built 26 age-prediction models for Brain Age based on different combinations of MRI modalities, using the Human Connectome Project in Aging (n=504, 36–100 years old). First, based on commonality analyses, we found a large overlap between Brain Age and chronological age: Brain Age could uniquely add only around 1.6% in explaining variation in fluid cognition over and above chronological age. Second, the age-prediction models that performed better at predicting chronological age did NOT necessarily create better Brain Age for capturing fluid cognition over and above chronological age. Instead, better-performing age-prediction models created Brain Age that overlapped larger with chronological age, up to around 29% out of 32%, in explaining fluid cognition. Third, Brain Age missed around 11% of the total variation in fluid cognition that could have been explained by the brain variation. That is, directly predicting fluid cognition from brain MRI data (instead of relying on Brain Age and chronological age) could lead to around a 1/3-time improvement of the total variation explained. Accordingly, we demonstrated the limited utility of Brain Age as a biomarker for fluid cognition and made some suggestions to ensure the utility of Brain Age in explaining fluid cognition and other phenotypes of interest.