Negative experiences and learning

The results, achieved over several years using highly sophisticated techniques that require a great deal of skill, could be summarised in a joking way, "there is now experimental evidence that not only humans but also experimental mice do not always learn from their unpleasant experiences", but it is better to be serious. Learning is of great importance thing. Whether we want to or not, we study for the rest of our lives - at least our brains strive to do so - and understanding the learning processes would be important evenly for healthy people and those who suffer from a wide range of neurological diseases.

It's no coincidence that one of the most studied brain areas within neuroscience research is the hippocampus, which is as important to learning and memory processes as it is to the brain, that the first Brain Prize for "research on the functional organization of neuronal circuits in the cerebral cortex" (Buzsáki, Freund, Somogyi, 2011) and the Nobel Prize for the discovery of place cells (2014, John O´Keefe, May-Britt Moser& Edvard I. Although the decade since then has not been short of discoveries, made possible by experimental techniques that seemed incredible only fifty years ago and were very quickly widely applied, we are still far from knowing enough about the workings of the "world's most complex computer", the human brain, to be able to avoid surprises in our research.

Balázs Hangya's latest work, published in Nature Communications, is related to memory and learning, but they did not conduct experiments on the hippocampus but studied deep brain inhibitory cells. Balázs Hangya, the team and research leader, answers questions.

- How did you conclude that the basal forebrain regulates a significant part of memory processes?

- It was suspected that the very first part of the basal forebrain, the medial septum, regulates memory, as it sends its fibers to the hippocampus, responsible for episodic memory, where it is important in generating the theta brainwave observed during exploratory behaviors. This fact has raised many questions, but let's consider the three most important ones. 

1. What does the septum do with memory? That it regulates the brainwave seems to be only half the answer, so we are still actively working on it. 

2. How does the septum control the theta wave, what is the mechanism? You could say that we are in a better position to answer this question because we published an article on it last year and the year before, but since we are talking about the same structure, we are working on answering the first and second questions in a coupled way.

3 Our third question builds on previous work by Panna Hegedüs (the first author of this article), who found that different cell types of the basal forebrain midbrain, Broca's diagonal bundle, are important in several aspects of learning.  We expected that there would be some consistency within the basal forebrain, so the question is this: If the anterior part of the basal forebrain is important for memory, what is the importance of its "continuation" that sends information to the cerebral cortex?

- Why did you start looking at the impact of negative experiences? Did you have any work hypotheses? 

- We didn't start with that. That's what we found out. We could have a hypothesis after Panna had done a lot of work to determine exactly when cells are active during a learning task. In this case, "exactly" means that this determination is made on cells identified by optogenetic labeling in chronic electrophysiological experiments. It is a difficult experiment. 

Parvalbumin fibres of the Broca's diagonal bundle (yellow) bind to the hippocampal calretinin calcium-binding protein cells (red). Credit: Panna Hegedüs

- All credit to Panna for the brilliant execution of the experiments! What became the hypothesis?

- For cholinergic cells, we had a specific hypothesis, as it was expected based on my paper published as a postdoc that these cells are important in predicting positive and negative outcomes.

Previous experiments have helped, in the rare case that the hypothesis has been proven true. Previously, we had a few experiments for the other main cell type here, parvalbumin (PV)-expressing GABAergic neurons, i.e. parvalbumin cells. The sound hypothesis was also arrived at only by technically difficult experiments. Fortunately, the need to apply a Pavlovian learning task was obvious. After the first experiment, we were already targeting "negative experiences".

 - It seems that more and more articles are dealing with the effects of negative experiences. Could this be due to the challenges of our lives and the consequent increase in the number of people with mental-psychiatric illnesses, especially depression? (Virtually one in five people are said to be affected by this condition for longer or shorter periods.)

- It may have something to do with it, yes. For example, depression may, according to one theory, be linked to an over-generalization of negative experiences, an over-extension of negative experiences to partially similar situations, which inhibits exploratory, reward-seeking behaviors. 

But it's also a factor that avoidance behaviors may be less hard to study because they involve very potent, mostly stimulatory brain pathways. Interestingly, we have demonstrated the importance of an inhibitory pathway, which does not only directly produce avoidance behavior but is more important in the association of negative values.

- Does the fact that a cell receives input from one or more cells involved in processing negative experiences make it clear that it is also involved in processing negative experiences?

- If not entirely (there are some oddities in the brain), it is highly likely, since direct inputs usually have a measurable effect on the functioning of the target cells. 

- People can react quite differently to the same event. How do mice behave?

- Mice tend to be very "optimistic generalizers", i.e. they extend rewarding experiences more strongly, so you could say, with a slight exaggeration, that everything is potentially rewarding for them until they find out otherwise. This can be evolutionarily adaptive as species compete for habitats. 

- How did the experiment work? 

- These mice perform a trained task to get a reward, such as a drop of water. However, in some cases, we blew air on the mouse's face for a moment, which was very unpleasant for it and it experienced it as a punishment. Both the probable reward and punishment were predicted by a single sound.

- When did you inhibit the inhibitory cells?

- The parvalbumin inhibitory cells of the anterior part of Broca's diagonal bundle were inhibited during the air blowing to avoid the association between the unpleasant experience and the predictive sound. As a result, the mice showed the same reward-seeking behavior after the reward and punishment predictive sounds. 

Starting inhibition after learning would not be effective, because the result would be reflected in the behavioral patterns performed before, rather than during, the inhibition period, in response to the predictive events.

- Inhibition of inhibition usually means stimulation, right?

- The inhibitory cell in adult individuals is mostly inhibitory, but often there is also inhibition release, where the inhibitory cell inhibits another inhibitory cell.  Inhibition release is likely an important mechanism in the action of the parvalbumin inhibitory cells we studied since we detected a high proportion of other inhibitory cells among their target cells. This is a common mechanism for long-range inhibitory cells, not an exception but in several cases the rule.

Thus, if inhibition of inhibition release is inhibited, activity in the target area, such as the cortex, is expected to decrease.

- How many cells and what big area of inhibitory cell activity had to be blocked for a behavioral change to occur?

- Relatively few cells and not over a large area. We also focused on the anterior part of the horizontal stem of the Broca's diagonal bundle. Going back to the previous question, it is probably precisely because of this that inhibiting a small number of inhibitory cells can be effective, because through inhibition of uninhibition (I hope it is tractable :D) we affect the activity of a relatively large number of cells. It is also known as an amplification, amplification mechanism.

 - Doesn't this complexity increase the vulnerability of the system? 

- I think the opposite. Complexity creates a lot of redundancy in the brain, and such a complex network can perform the same task in multiple ways. 

- Let's end our little talk about negative experiences with something positive!

What do you respect as the greatest discoveries of the 21st century, now almost a quarter of a century old? 

- Fortunately, there are many to choose from. If we take neurophysiology, one of my favorite examples is the understanding of the function of the brain stem cells to a level that has enabled deep brain stimulation therapy for Parkinson's disease (e.g. the work of Alim Louis Benabid and Mahlon DeLong). I would highlight the exploration of the role of immunological processes in neurodegenerative diseases and, although this is still an open question, there may be great potential for new oscillatory therapies for Alzheimer's disease.