Using a two-photon imaging microscope and a wealth of genetic tools, researchers from Cold Spring Harbor Laboratory (CSHL), Columbia University, University College London, and Flatiron Institute found that neural networks become more focused as mice got better at performing a trained task. They used the data to construct computational models that can inform their understanding of the neuroscience behind decision-making.

“We recorded the activity from hundreds of neurons all at the same time, and studied what the neurons did over learning,” said CSHL Associate Professor Anne Churchland. “Nobody really knew how animals or humans learn the structure of a task and how the neural activity supports that.”

The team, including Farzaneh Najafi, the first author on the study and a postdoctoral fellow in Churchland’s lab, started by training mice on perceptual decision-making tasks. The mice received multisensory stimuli in the form of a sequence of clicks and flashes that were presented together. Their job was to tell researchers whether those are happening at a high or low rate by licking one of three waterspouts in front of them.

They licked the middle spout to start the trial, one side to report a high-rate decision and the other side for a low-rate decision. When the mice made the correct decision, they received a reward.

“Most decision-making studies focused on the period where the animals are really experts. But we were able to see how they arrive at the state by measuring the neurons in their brain all the way through learning,” said Churchland, the senior author on the study. “We found that in all the animals, their learning occurs gradually over about four weeks. And we found that what supports learning is activity changes in a whole bunch of neurons.”

The neurons, the team discovered, became more selective in responding to an activity associated with a particular task. The also started reacting faster and more immediately.

“They’ll respond really strongly in advance of one choice and much less so in advance of the other choice,” Churchland said.

When the animals are just beginning to learn, the neurons don’t respond until around the time it makes the choice. But as the animal gains expertise, the neurons respond much further in advance.

“We can kind of read the animal’s mind in a way, we can predict what the animal is going to do before he does it,” Churchland said. “When you’re a novice at something your brain is doing all different things, so you have neurons engaged in all different things. But then when you’re an expert, you hone in on exactly what you’re going to do and we can pick up that activity.”

The researchers decoded neural activity by training a small artificial network called the ‘Linear Support Vector Machine’ using machine learning algorithms. It collects performance data from multiple trials and combines it with the activity of all the neurons, weighing them to make a guess about what the animal’s going to do. As the animal gets better at the task, its neural networks get more refined, precise and specific. The researchers are able to mirror that onto the artificial network, which can then predict the animal’s decision with about 90 percent accuracy.

The learning models also offer another way of looking at specific types of neurons in the brain involved in cognition, like excitatory and inhibitory neurons, which trigger positive and negative changes, respectively. In this study, published in Neuron (Cell Press), the team found that the inhibitory neurons are part of very selective sub-networks in the brain, and they’re strongly selective for the choice that the animal’s going to make.

These neurons are part of a biophysical model that helps researchers understand how decision making works. As researchers refine these models, they’re able to make more sense of how cognition informs behavior.

“We’ve learned a lot about perceptual decision-making-the decisions that a subject would get right and wrong, how long it takes to make those decisions, what the neural activity would look like during decision-making-by making different kinds of models that make really concrete predictions,” Churchland said. “Now we can understand, hopefully better, why these very selective sub-networks are there, how they help us make better decisions, and how they are wired up during learning.”

In a paper published in Science Advances this week, physicists Professor Sum Tze Chien from NTU and Professor Maxim Pshenichnikov from UG used extremely fast lasers to observe how an energy barrier forms when perovskite is joined with a material that extracts the electrical charges to make a solar cell.

Conventionally, a solar cell absorbs sunlight and converts it to electrical charge. During this process, the light particles have more energy than needed to generate the electrical charges in the solar cells.

This excess energy gives rise to what are called “hot” charges, which lose their excess energy very fast as heat (within one picosecond), leaving only “cold” charges available for electrical power generation.

This energy loss is why conventional solar cells have a theoretical limit of 33 per cent for power conversion efficiency. The best perovskite solar cells so far have exhibited 25 per cent efficiency, almost on a par with the best performing silicon solar cells.

Scientists believe that if “hot” charges could be extracted fast enough, then together with the harvested “cold” charges, it could lead to a “hot carrier” solar cell with a theoretical efficiency of up to 66 per cent.

The key to extracting these hot charges quickly enough lies in the selection of the correct ‘extraction’ material to bond with the perovskite. Prof Sum’s team has now devised a way to measure which are the best extraction materials.

Prof Sum, the Associate Chair (Research) at NTU’s School of Physical and Mathematical Sciences, said “Our latest findings show how ‘hot’ these charges have to be, in order to cross over the energy barrier without being wasted as heat. This highlights the need for better pairing of ‘extraction’ materials with perovskites if we want to lower this energy barrier for more efficient solar cells.”

Perovskite solar cells’ primary advantage over silicon solar cells is that they are cheap and easy to manufacture using common chemistry laboratory supplies and do not need silicon’s costly and energy intensive manufacturing processes.

Prof Sum and his collaborators previously published in Science their discovery that “hot” charges in perovskites lose their excess energies more slowly than in other semiconductors. The team subsequently slowed this energy loss further using nano-sized perovskites, making it easier to extract the hot charges as electricity.

In their latest experiments, the NTU and UG scientists ‘watched’ the solar cells at work using femtosecond pulsed lasers that can measure processes that occur roughly 100 billion times faster than a camera flash. The scientists studied the behaviour of the “hot” charges that are generated and how they moved through the perovskite into the extractor material without losing their excess energy as heat.

Prof Pshenichnikov said, “Such high efficiency solar cells could mean the possibility of increasing the energy supply from solar panels without the need for more surface area.”

Giving an independent comment on the research, Dr Henk Bolink from the Institut de Ciència Molecular (ICMol) of the Cientific Parc of the University of Valencia, said besides a suitable light absorbing layer, solar cells also need charge extraction layers that selectively extract either electrons or holes to the two terminals of the cell.

“It is currently unclear what the charge extraction interface composition/property should be, to allow for the extraction of both the “hot” and “cold” charges,” said Dr Bolink, who was not involved in the study.

“In their recent work, Prof Sum and Prof Pshenichnikov shed light on this crucial puzzle by demonstrating a method that allows for the identification of the suitability of these charge extraction layers.”

The study — “Trends and Rapidity of Dose Tapering Among Patients Prescribed Long-term Opioid Therapy, 2008-2017” — is published in JAMA Network Open. The results also will be presented at the Nov. 16-19 North American Primary Care Research Group meeting in Toronto.

“Tapering plans should be based on the needs and histories of each patient and adjusted as needed to avoid adverse outcomes,” said study author Alicia Agnoli, assistant professor of family and community medicine. “Unfortunately, a lot of tapering occurs due to policy pressures and a rush to get doses below a specific and sometimes arbitrary threshold. That approach can be detrimental in the long run.”

In 2016, the U.S. Centers for Disease Control and Prevention (CDC) recommended dose tapering, or a slow reduction in prescription opioid doses over time, if the risks of continuing opioids outweigh the benefits. That point in time is usually when a patient is taking 90 morphine milligram equivalents — or MMEs — each day, and that dose is no longer reducing pain or improving daily functions. The CDC advises a slow decrease of 10% MMEs per month.

The study team set out to examine trends in opioid dose tapering and if tapering rates were consistent with CDC recommendations.

“We wanted to understand how often opioid dose tapering happens, how rapidly patients’ doses were being reduced when tapering, and which patients were more likely to have doses tapered,” said lead author Joshua Fenton, professor of family and community medicine.

Tapering faster than recommended

Fenton and Agnoli evaluated medical and pharmacy claims and enrollment records for more than 100,000 commercial insurance and Medicare Advantage enrollees, representing a diverse mixture of ages, races, ethnicities and locations across the U.S. They focused on individuals whose opioid doses were stable for at least a year and identified tapering patients as those with a 15% or more reduction in daily MMEs during a seven-month follow-up period.

They found that dose tapering became more common throughout the study period of 2008-2017, with the biggest jump following the CDC’s 2016 prescribing guidelines. Tapering increased from 10.5% to 13.7% from 2008 to 2015, and from 16.2% to 22.4% from 2016 to 2017. Tapering was much more common in patients prescribed higher opioid dosages.

They also found that the rate of dose reduction often was well beyond the CDC’s recommendation of 10% per month. The average reduction overall was 27.6% per month. Nearly 20% of patients tapered at a rate of 40% per month, and 5% tapered at a rate faster than 60% per month.

The 2016 policy could have been misinterpreted, leading many prescribers and health systems to insist on faster-than-recommended tapering, according to Agnoli.

“There is definitely a lot of pressure to reduce opioid use among patients, but there also is a need for more training and guidance for prescribers on how to help them safely do so,” Agnoli said.

Women and tapering

Fenton and Agnoli also identified patient variables associated with tapering and uncovered an interesting difference in tapering rates based on sex.

While men have much higher rates of opioid use disorder and adverse outcomes related to opioids, women were more likely than men to have their opioid doses tapered.

“We think this has a lot to do with the gender dynamics of pain management and the physician-patient relationship,” Agnoli said. “How women experience pain and discuss pain with their physicians is perhaps very different than men. There also could be some sex bias in terms of the patients that physicians choose to initiate conversations with about dose reduction.”

Minimizing tapering risks

The researchers hope to build on this work to inform best practices for safe decision-making around dose reduction for all patients prescribed opioids.

“Ultimately, we want to clarify the effects of tapering on patients and how to help them taper to maximize benefits and minimize risks,” Fenton said. “We expect this line of research will have important implications for how physicians manage and monitor patients who are undergoing opioid tapering.”

Collaborators on this research were Guibo Xing, Daniel Tancredi, Anthony Jerant and Elizabeth Magnan of UC Davis Health, and Lillian Hang and Aylin Altan of OptumLabs in Minneapolis.

Their study was supported by UC Davis Health and OptumLabs and is available online.