Content at a glance : Dopamine is an important neurotransmitter in the nervous system, which is closely related to movement, memory and reward system. search. However, accurate quantitative analysis of dopamine is still elusive. With the help of machine learning, Markita P. Landry's research group at the University of California, Berkeley (UCB) conducted a quantitative analysis of the release amount and location of dopamine, bringing us one step closer to the happiness code.
Keywords : machine learning reinforcement learning dopamine
Author｜Xuecai
Editor｜Sanyang
This article was first published on the HyperAI Super Neural WeChat public platform.
We are often asked the question "Are you happy?". After reviewing our recent living conditions, we may be able to make a relatively satisfactory answer. However, answering another question about happiness, how happy are you, is not so easy.
We can make a relatively accurate judgment of happiness, but it is difficult to conduct a quantitative analysis of happiness, and we can only use some degree adverbs to make a rough evaluation.
But from a physiological point of view, the degree of happiness can be judged by the level of hormones in the human body, and one of the important hormones is dopamine .
Figure 1: The four pleasure hormones from left to right are dopamine, endorphins, oxytocin, and serotonin
Dopamine is an important neurotransmitter in the nervous system, responsible for transmitting messages between cells . Dopamine is the messenger of pleasure. When we see something pleasurable, the brain releases dopamine, prompting us to pursue it. Therefore, a neural circuit controlled by dopaminergic neurons is also called a reward circuit, which is closely related to learning, memory and addictive behavior.
Although people have a relatively clear understanding of the chemical structure, distribution area and physiological functions of dopamine, the mechanism of action of dopamine at the cellular and molecular levels is still not well understood, let alone the role of dopamine in neural circuits. Carry out accurate quantitative analysis .
"Quantified" happiness: AI deciphers dopamine code
In 1997, Schultz et al. proposed a possible operating mechanism of the reward circuit - the reward prediction error hypothesis . This hypothesis holds that dopaminergic neurons will adjust the release of dopamine according to the error between the expected reward and the actual reward, and then adjust people's motivation to pursue something.
In 2020, DeepMind discovered in the brain that different neurons have different reward expectations for the same stimulus . In other words, there are relatively optimistic neurons and pessimistic neurons in the brain. Faced with the same half glass of water, optimistic neurons will think that with half a glass left, we have a bright future. The pessimistic neurons will think that there is only half a glass of water left, and we are going to die of thirst. Moreover, further studies have shown that the distribution of neurons' expected rewards is basically consistent with the distribution of actual rewards.
Figure 2: Expected reward (blue) and actual reward (gray) for a neuron
With the help of AI, the understanding of the neural mechanism of the reward circuit is accelerating .
In 2021, Erin S. Calipar's research group at Vanderbilt University (Vandy) in the United States monitored changes in dopamine content in organisms, and used support vector machines (SVM) to predict the behavior of organisms. At the same time, based on the experimental results, the research team A new model for the regulation of physiological activity by dopamine is proposed.
Recently, AI's interpretation of dopamine has been improved. With the help of machine learning, Markita P. Landry's research group at the University of California, Berkeley (UCB), conducted a quantitative analysis of the release of dopamine and the release of brain regions, providing new ideas for the study of neuroimaging and neural circuits .
Related research has been published in "ACS Chemical Neuroscience", titled "Identifying Neural Signatures of Dopamine Signaling with Machine Learning".
Figure 3: The research results have been published in "ACS Chemical Neuroscience"
Paper address: https://pubs.acs.org/doi/full/10.1021/acschemneuro.3c00001
This study mainly addressed two questions:
1. Distinguish the amount of dopamine released under different stimuli (0.1 mA and 0.3 mA current stimulation);
2. To determine the release of dopamine in the brain area (DLS of dorsolateral striatum and DMS of dorsomedial striatum).
First, they labeled dopamine with near infrared catecholamine nanosensors (nIRCat, near infrared catecholamine nanosensors). After labeling, under an infrared microscope, dopamine will emit fluorescence, and the fluorescence intensity is positively correlated with the concentration of dopamine . When electrical stimulation is applied to the brain, dopamine is released and then recycled. This process will leave a fluorescence intensity curve under the infrared microscope. By quantifying the fluorescence curve, 8 statistical features can be obtained, such as the average fluorescence intensity, the number of dopamine release sites (ROI, regions of interests), etc., and 2 A temporal feature, including the duration of fluorescence intensity above and below 2 standard deviations. These feature values can be used in the training of machine learning models.
Figure 4: nIRCat labeling results for dopamine
A : Fluorescence results observed before and after current stimulation
B : Fluorescence intensity curves before and after current stimulation
The researchers trained and analyzed two models, a support vector machine (SVM) and a random forest model (RF), respectively .
The SVM model can classify the results into two categories based on complex nonlinear features, and apply the trained boundary conditions to the test data. The RF model consists of multiple decision trees, and the decisions made by each decision tree are finally collated together to obtain the final output.
The RF model can fully interpret the variables in the results to ensure accurate predictions. By randomly selecting data and features, the sensitivity of the decision tree model to the original training data is reduced, and the difference between decision trees is improved.
The amount of training data required by the two models is small, and the results can be divided into two categories, which matches the purpose of this study.
Figure 5: Workflow for machine learning
Data Set A and Data Set B : respectively represent the dopamine release concentration of different current stimulation or different brain regions
After the training of the two models is completed, the fluorescence intensity curve obtained under different current stimulation is used as input, and the model can judge the intensity of the stimulation received and the brain area released by dopamine.
Figure 6: Judgment results of machine learning for different stimulus intensities
Figure A : Judgment results for 4-week-old mice
Figure B : Judgment results for 8.5-week-old mice
Figure C : Judgment results for 12-week-old mice
It can be seen from the results that as the age of the mice increases, the accuracy of the two models for judging the intensity of the stimulus increases continuously . This is mainly because, as the age of the mice increases, the hormone levels in the body gradually stabilize and are easy to predict. On 12-week-old mice, the accuracy rate of the RF model for judging the stimulus intensity can reach 0.832.
Figure 7: Under 0.3 mA current stimulation, the accuracy of machine learning on dopamine-releasing brain regions (left) and the importance of different features on the accuracy of judgment (right)
A&B : Judgment results for 4-week-old mice
C&D : Judgment results for 8.5-week-old mice
E&F : Judgment results for 12-week-old mice
It can be seen from the figure that, similar to the results of stimulus intensity, machine learning has the highest judgment accuracy rate on 12-week-old mice, up to 0.708. At the same time, different input features will also affect the judgment accuracy of the model. Among the different characteristic parameters, ROI is the most important for the judgment accuracy of the model .
Through machine learning, the researchers broke the shackles of traditional data analysis, selected a large number of feature variables, and improved the judgment accuracy of the model through the characteristic ROI that traditional data analysis ignores. In addition, this model can also be extended to neural circuits other than dopamine, providing new ideas for the study of neuroimaging and neural mechanisms .
Dopamine: The Double-Edged Sword of Pleasure and Loss
Dopamine brings us pleasurable feelings and drives us to seek out pleasurable things. Whether it's delicious food, gorgeous scenery, proper exercise or active social interaction, they all contribute to the release of dopamine, which helps us maintain a good mood . Because of this, dopamine can also be used as a marketing tool for merchants. From the exquisitely packaged "Dopamine Catering" to the "Dopamine Outfit" sweeping social media, bright colors not only embellish people's lives, but also lighten people's moods.
Figure 8: The dopamine outfit of the UP master "Kangkang and Grandpa"
However, after being happy, the dopamine level in the body will temporarily drop below the normal level, which will bring depression instead . After long-term and frequent secretion of dopamine, the human body's perception of happiness will become dull, making it difficult for people to appreciate the beauty of life, and it is easier to become lost. Therefore, some people also put forward the concept of "dopamine withdrawal". By adjusting work and rest, controlling entertainment time, staying away from social media, etc., the release of dopamine in the body can be controlled, so as to return to life and experience real happiness.
Whether it's "Dopamine Dressing" or "Dopamine Withdrawal", everyone is pursuing the beauty in life and making themselves happy. Although the two theories have certain physiological basis, the actual effect remains to be studied . With the help of AI, researchers are also constantly digging out the mechanism behind neural activity and exploring the mysteries of dopamine. I believe that one day, when asked "how happy are you", people will be able to answer without hesitation, 100%.
This article was first published on the HyperAI Super Neural WeChat public platform .
Reference article:
[1]https://www.nature.com/articles/s41586-019-1924-6#additional-information
[2]https://www.sciencedirect.com/science/article/pii/S096098222101188X
[3]https://www.science.org/doi/10.1126/science.275.5306.1593
[4]https://prezi.com/gxadjg6gz7li/nicotine-and-the-brain-reward-system/
[5]https://youtu.be/v6VJ2RO66Ag