Saturday quote: The sound of music, sneaky birds, better training for LLM. Plus: Diversity improves research

In the small fishing village where I grew up, we didn’t have much. But we helped our neighbors, raised our children to respect the sea, and adopted an inclusive scientific methodology across gender, race, and gender among study participants that enriched our research findings.

This week we reported on a study that changed previous brain research findings by including both men and women. The researchers also report that cuckoos may have a more important ecological function than previously known. Plus a new LLM training mod and insight into how people distinguish music from speech.

Jerk birds increase biodiversity

Many cuckoos are parasites that lay their eggs in the nests of other birds; when they hatch, the cuckoo chicks kick the other eggs out of the nest so their foster parents aren’t distracted by their own chicks and feeding their chick a parasite for weeks they could spend raising their own family.

The warblers have adapted to reject odd-looking offspring, so bronze cuckoo chicks have evolved to resemble their host parents. A new study by scientists from the University of Cambridge suggests that this current evolution is leading to new species of cuckoos and thus to biodiversity in general.

The researchers conducted an extensive analysis of all cuckoo species and found that the lines that are the most costly in terms of resources and breeding time for their hosts have adapted to look as much like their host birds as possible. Dr. Clare Holleley from CSIRO says: “This finding is significant in evolutionary biology and shows that coevolution between interacting species increases biodiversity by driving speciation.”

Scientists have found that women are also useful

When it comes to gender diversity, much scientific research is stuck in the 1950s, with researchers predominantly recruiting men as subjects for studies and experiments with the assumption that the findings are equally applicable to women. Now, Weill Cornell Medicine researchers report the first evidence that astrocyte receptors in the brain produce opposite effects on cognitive function between male and female clinical models, suggesting that astrocytes contribute to sex-specific brain mechanisms. Previous studies of astrocytes on cognitive function considered only men, leading to the widespread assumption that the mechanism was the same between the sexes.

Gender differences are known for many neurological conditions, including schizophrenia, stroke and dementia. Focusing on mGluR3, a glutamate receptor in astrocytes and a gene highly altered in dementia, the research team selectively manipulated astrocytes in animal models to examine the effects of mGluR3 on learning, memory, and cognitive and behavioral outcomes.

In women, increasing levels of mGluR3 improved memory in older women and reduced impaired memory in younger women. In men, reducing mGluR3 improved memory and increasing levels had no effects. “Therapeutics affecting astrocytic receptors may cause sex-specific cognitive effects in part because of the different roles of astrocytes in males and females,” said Dr. Anna Orr.

Better pedagogy for LLM

By training LLMs with a new modality similar to language training in the human brain, researchers at Hong Kong Polytechnic University have developed artificial intelligence models that function similarly to humans. Current LLM training is designed around contextual word prediction. Instead, the researchers examined next-sentence prediction tasks, simulating discourse-level comprehension in the human brain, to assess the coherence of a pair of sentences.

They trained two models, one with the NSP enhancement and one without. They were also trained in standard word prediction. The researchers compared the patterns from the models with fMRI data collected from people reading either connected or disconnected sentences. The model with NSP training provided a better fit to human brain activity than the LLM without NSP training. The researchers hope to create new training modalities for LLMs that do not rely solely on a huge corpus of text to produce intelligent results.

Music, speech different

Humans are able to distinguish the sounds of music from speech and focus narrowly on speech in an environment with music. An international team of researchers has mapped the process in a study that could improve therapy programs that use music to help people with aphasia regain the ability to speak.

The researchers played audio clips to study participants who were told they would hear speech or music masked by noise and asked them to distinguish between them. While the participants sorted through hundreds of clips, the researchers observed to what extent the features of speed and regularity influenced their judgments.

According to scientists, the brain uses simple acoustic parameters to distinguish speech from music. They found that slower, steadier audio clips with random noise sounded more like music, while fast, irregular clips sounded more like speech. Specifically, speech is two to three times faster than most music. Moreover, changes in volume (or amplitude modulation, if you’re trying to impress a reviewer) are stable in music, whereas the volume of human speech changes frequently.