Australasian Science: Australia's authority on science since 1938

The Light Bulb Moment for Brain Development

A zebrafish head viewed from front-on showing neurons in a slice of its brain labelled with a fluorescent calcium indicator.

A zebrafish head viewed from front-on showing neurons in a slice of its brain labelled with a fluorescent calcium indicator.

By Geoffrey Goodhill & Lilach Avitan

Some elegant experiments in zebrafish have revealed how sensory experience during infancy can have long-lasting effects on the brain.

The brain develops through a combination of innate and environmental factors. The initial patterns of wiring are specified genetically, but this provides only a rough starting framework. From there, neural activity is required to refine these patterns. Using the zebrafish as a model, we have now discovered more about how this process works.

Neural activity in the brain is generated in two ways. Most obviously it is caused by sensory inputs such as sights, sounds and smells. However, it can also be generated spontaneously within the brain without an external stimulus. It’s as if the brain is internally rehearsing the kind of input patterns it expects to encounter from sensory stimuli.

How does this spontaneous activity change over development, and how is it affected by the sensory environment? It’s obviously hard to study this in humans, but zebrafish offer an attractive alternative.

Unlike humans, zebrafish grow very quickly. Within only 5 days of the egg being fertilised, when they are only a few millimetres long, they are mature enough to begin hunting fast-swimming paramecia.

Larval zebrafish are also transparent, and their neurons can be genetically labelled with a fluorescent calcium indicator. When a neuron fires, calcium flows into it so that labelled neurons glow brighter when they are active.

We can therefore study brain activity in the zebrafish by simply embedding the fish in agarose under a microscope and watching the neurons glowing. This makes it possible to observe hundreds or even thousands of neurons simultaneously.

Our focus has been on a brain region called the optic tectum. This is the main part of the zebrafish brain that processes visual information coming from the eyes. While the tectum is active in response to visual stimulation, during the early life of the zebrafish it also displays large amounts of spontaneous activity.

We examined this spontaneous activity every day from 4–9 days post-fertilisation (dpf). Our first finding was that the amount of activity changed with age. From 4–5 dpf, spontaneous activity increased in frequency. However, it then declined, until by 8 dpf the frequency was back to 4 dpf levels.

When recording from hundreds of neurons simultaneously, there is much more to be discovered about their activity patterns than just frequency. We therefore looked at the activity correlations between neurons, and this told us which pairs of neurons tended to fire together.

This allowed us to extract the “functional connectivity” of the tectum (i.e. which neurons might be connected) based on the fact they had correlated patterns of activity. Just like the frequency of activity, we found that this functional connectivity peaked at 5 dpf and then declined again.

However, we also noticed that large groups of neurons tended to fire together as “neural assemblies”. To determine which neurons were in which groups, we turned to “community detection” algorithms recently proposed in an area of mathematics called graph theory. Inspired in part by the huge amounts of data now available about online social networks, these algorithms determine the form of “communities” in such networks. Treating neural firing just like a social network, we used these algorithms to discover that the number of communities, and the number of neurons within each community, also peaked at around 5 dpf.

It thus appears that 5 dpf is a particularly important moment in the development of the zebrafish’s optic tectum. This is exactly when they start to hunt for food.

Sensitivity to Visual Stimulation

Is the development of this spontaneous activity purely driven by an intrinsic genetic program, or does it also depend on the visual stimulation the zebrafish receives early in life? To answer this question we performed two different kinds of manipulation of the visual input. First, we raised one group of fish until 6 dpf in the dark. Second, we raised another group of fish until 6 dpf on a normal light-dark cycle but in a “featureless” environment where they saw no visual contours, just diffuse light. We then imaged spontaneous activity in the tectum as before.

Both manipulations caused changes to the structure of spontaneous activity. Dark-rearing caused a loss of functional connectivity and fewer neural assemblies. In contrast, functional connectivity increased after featureless rearing (i.e. it did not refine, as occurs in normally reared fish).

Thus normal visual inputs are essential for the appropriate refinement of activity patterns in the optic tectum.

Behavioural Consequences

While we showed that dark-rearing during early life changed subsequent brain activity, a key question remaining was whether this also had an effect on the behaviour of the fish.

We investigated this using a very simple assay for the ability of the fish to hunt paramecia. This is a challenging task, since paramecia are single-cell creatures that move quite quickly. We know from previous work that zebrafish catch paramecia primarily through their visual sense.

At 6 dpf each fish was placed in a dish in the light with 50 paramecia – the first time they had seen paramecia or tried to catch their own food. After 2 hours we counted how many paramecia remained, telling us how many had been eaten.

Fish that had received normal visual inputs before 6 dpf ate about half of the paramecia. However, fish reared in the dark until that point ate hardly any, even though they were hunting under exactly the same conditions as the normal fish.

Remarkably, this deficit persisted to 9 dpf even though both sets of fish received the same normal visual input from 6–9 dpf. Thus, dark-rearing causes a profound and long-lasting change in the ability of the fish to catch prey using visual cues.

Critical Periods

A large body of previous research has shown that the mammalian brain exhibits critical periods in its development. These are relatively brief moments when particular parts of the brain are unusually sensitive to the inputs they receive. Unless it gets the right inputs at those times it will not wire up properly, even if the input is later corrected.

A seminal example is the 1960s discovery that briefly blocking input to a cat’s eye during the critical period causes a permanent vision deficit in that eye, while blocking visual input after the critical period has no long-term effect. These findings in cats changed the way that children with vision problems early in life are treated by doctors.

It was previously thought that the zebrafish brain developed according to a more rigid genetic program. However, our work shows that, like mammals, their brains are also sensitive to environmental inputs early in life. The relative ease with which neural activity can be imaged in the young zebrafish brain thus provides an exciting opportunity to study brain plasticity in more detail than is possible with cats or humans.

The Future

There are many questions we would now like to address. For instance, is the plasticity we have observed really a “critical period” phenomenon and, if so, how long does this critical period last? Are the changes in brain wiring observed with altered input reversible? If so, what are the most effective kinds of visual input to achieve this?

An essential part of our approach to this work has been the application of advanced mathematical and statistical methods to extract the maximum amount of information from our data. Indeed, several people in the team originally came from backgrounds in mathematics, physics or computer science. We believe this kind of interdisciplinary approach will be increasingly important for unravelling the highly complex rules governing brain development and function.


Geoffrey Goodhill is Professor of Neuroscience and Mathematics at The University of Queensland. Lilach Avitan is a Research Fellow at the Queensland Brain Institut, The University of Queensland.