CP v2

Category: Relationships

  • Study Tracks Overimitation in Infants: Copying Starts Early, but In-Group Bias Comes Later

    Infants begin copying unnecessary actions well before age two, but that early tendency does not yet appear tied to choosing people who seem more similar to them. That is the central finding of a new Concordia University study published in Frontiers in Developmental Psychology.

    Overimitation refers to copying steps that are irrelevant to achieving a goal, such as repeating an extra action when opening a box to reach a toy. Researchers have long linked the behavior in older children to social affiliation, but evidence in children under two has been limited.

    What the new study tested

    The team observed 73 children aged 16 to 21 months, with an average age just over 18 months. Each child completed four tasks designed to measure different types of imitation and a separate test for in-group preference.

    In the overimitation task, an adult demonstrated three steps to open a box, including one action that did not help retrieve the toy. Other tasks measured memory-based copying and the ability to infer an adult’s intended goal when the adult appeared to fail at completing an action.

    Low overimitation, no in-group pull

    The researchers found low levels of overimitation at this age, and children’s performance was not driven by in-group preference. In the in-group task, children chose between objects offered by a woman and a robot, and their choices did not predict overimitation.

    By contrast, two other imitation measures showed a clear relationship: elicited imitation, often used to assess early memory, and imitation of unfulfilled intentions. This pattern suggests that at 16 to 21 months, imitation is more closely aligned with developing cognition and recall than with group-based social motivations.

    Why it matters for parents and educators

    The authors argue that the social reasons often proposed for overimitation may emerge later in development, as children learn more about group membership. They point to related work indicating that by around preschool age, overimitation can align more with preferences for similar peers.

    The findings also serve as a reminder that very young children may copy both helpful and unnecessary behaviors from adults. The researchers say this has implications for modeling actions in homes and classrooms where early learning is strongly shaped by observation.

  • MIT’s ComMAND gene circuit could make gene therapy dosing safer and more predictable

    Gene therapy has long promised one-time treatments for disorders caused by a missing or faulty gene, but controlling how strongly a delivered gene turns on in cells remains a major hurdle. Too little expression can leave a therapy ineffective, while too much can raise toxicity and other safety risks.

    Engineers at MIT report a compact gene-control design that aims to keep expression within a targeted range, even when cells receive different numbers of gene copies. The work, published in Cell Systems, centers on a circuit the team calls ComMAND, short for Compact microRNA-mediated attenuator of noise and dosage.

    A built-in brake for expression

    Many gene therapies rely on viral vectors such as adeno-associated virus or lentivirus to deliver therapeutic DNA. But uptake varies widely from cell to cell, which can create large swings in how much protein a new gene produces.

    ComMAND uses a control strategy known as an incoherent feedforward loop, pairing gene activation with a simultaneous suppressor signal. In this design, the therapeutic gene also produces a microRNA that dampens its own translation, acting as an internal counterweight.

    Compact design fits common vectors

    The researchers engineered the microRNA sequence inside an intron within the therapeutic gene, so both the gene’s messenger RNA and the suppressing microRNA are produced together. That single-transcript setup is intended to smooth out variability when delivery levels differ across cells.

    Because the circuit can be controlled with one promoter, the team says expression can be tuned by selecting promoters of different strengths. The compact architecture is also designed to fit within a single delivery vehicle, which could simplify manufacturing and development.

    Early results across multiple cell types

    In human cells, the team demonstrated ComMAND with genes linked to Friedreich’s ataxia and Fragile X syndrome, aiming to keep expression closer to desired levels. They reported gene output around eight times typical healthy levels in their tests, compared with more than 50 times without the circuit.

    The approach was also evaluated in rat neurons, mouse fibroblasts, and human T-cells using a fluorescent reporter to measure expression. The researchers say the next step is to test whether this tighter control can restore function and improve disease signs in cultured systems and animal models.

    The authors note that many candidate conditions are rare, making it difficult to run large studies and optimize dosing. They argue that more predictable, tunable gene circuits could lower development barriers for therapies targeting small patient populations.

  • Hong Kong PolyU researchers unveil AI tool to score large language model personality, with implications for business compliance and education

    Large language models have become a default interface for many AI products, but researchers still struggle to describe their behaviour in a consistent, measurable way. A team at The Hong Kong Polytechnic University says it has built a system that aims to quantify an LLM’s personality based on linguistic output.

    The tool, called Language Model Linguistic Personality Assessment, or LMLPA, is designed to translate model responses into numerical scores tied to personality traits. The researchers describe it as a step toward making model behaviour easier to compare across systems and deployments.

    How the LMLPA system works

    LMLPA combines two components: an adapted version of the Big Five Inventory and an AI rater that grades the model’s answers. The process focuses on patterns such as wording, style and other language features found in generated text.

    By using a standardised questionnaire structure, the approach attempts to bring more consistency to assessments that often rely on subjective impressions. The researchers say the outcome is a data-driven profile that can be tracked and tested across different prompts and settings.

    Why personality metrics could matter

    Developers and organisations increasingly want AI assistants to behave predictably in sensitive contexts, including classrooms, customer support and internal decision workflows. The team argues that quantifying communication tendencies could help tailor a model’s tone and interaction style to a specific use case.

    The researchers also point to potential value in governance and oversight, where firms are under pressure to document how AI tools behave and how risks are managed. They suggest that structured behavioural metrics could complement existing evaluation methods focused on accuracy and safety.

    From research to compliance applications

    PolyU said the work has also informed a business compliance platform that uses natural language processing to analyse large volumes of reports and other text. In that context, automation is intended to speed up data collection, analysis and insight generation for reporting tasks.

    The study, led by Prof. Lik-Hang Lee of PolyU’s Department of Industrial and Systems Engineering, was published in the journal Computational Linguistics. The researchers position LMLPA as part of a broader effort to align AI systems with human values and practical operational needs.

  • Tinnitus severity may be measurable at last: Study links pupil dilation and facial micro-movements to distress levels

    Researchers at Mass General Brigham say they have identified potential objective biomarkers for tinnitus severity by tracking pupil dilation and subtle, involuntary facial movements while people listen to everyday sounds. The work, published in Science Translational Medicine, aims to address a long-standing problem in tinnitus research: severity is typically judged by questionnaires rather than physiological measures.

    Tinnitus is commonly described as persistent phantom sound, such as ringing, buzzing, or clicking, and it is widespread in the general population. For many people it is a manageable nuisance, but a smaller group experiences debilitating distress that can disrupt sleep, concentration, and mental health.

    Signals tied to threat response

    The team focused on the sympathetic nervous system, which governs the body’s fight, flight, or freeze response, to look for outward indicators of distress. They examined whether tinnitus-related distress might be reflected in arousal signals that are visible in the eyes and face.

    To test the idea, researchers recruited 97 participants with normal hearing, including 47 with varying levels of tinnitus and sound sensitivity, and 50 control volunteers. Participants listened to pleasant, neutral, and unpleasant sounds while being recorded on video and monitored for pupil changes.

    AI helps detect tiny facial changes

    Using AI-powered video analysis, the study detected rapid micro-movements in areas such as the cheeks, eyebrows, and nostrils, and found they were associated with self-reported tinnitus distress. When these facial signals were combined with pupil dilation data, the model’s ability to predict severity improved.

    People with severe tinnitus showed unusually large pupil dilation across all sound types, while their facial responses were more muted. By contrast, controls and participants with less bothersome tinnitus tended to show stronger pupil and facial reactions mainly to the most unpleasant sounds.

    Why this could matter for trials

    The researchers argue that objective readouts could make placebo-controlled studies easier to design and interpret, helping the field evaluate treatments more rigorously. They also suggest the approach could potentially be adapted to more accessible, clinic-friendly tools if validated further.

    The team noted key limitations, including the need to exclude many people who often have complex tinnitus, such as those with hearing loss, older age, or significant mental health comorbidities. Future studies are expected to test whether the biomarkers hold up in broader, higher-risk populations and in real-world clinical settings.

    Researchers involved in the work say they are now exploring how these biomarkers could support therapy development, including approaches that pair neural stimulation with software-based treatment environments. The broader goal is to measure not just the sound people perceive, but the distress response that makes tinnitus disabling for some patients.

  • Explainable AI tool CANYA decodes protein aggregation patterns, offering new clues for amyloid diseases and drug manufacturing

    An AI tool has made a step forward in translating the language proteins use to dictate whether they form sticky clumps similar to those linked to Alzheimer’s Disease and around fifty other types of human disease. In a departure from typical “black-box” AI models, the new tool, CANYA, was designed to be able to explain its decisions, revealing the specific chemical patterns that drive or prevent harmful protein folding.

    The discovery, published today in the journal Science Advances, was possible thanks to the largest-ever dataset on protein aggregation created to date. The study gives new insights about the molecular mechanisms underpinning sticky proteins, which are linked to diseases affecting half a billion people worldwide.

    Protein clumping, or amyloid aggregation, is a health hazard that disrupts normal cell function. When certain patches in proteins stick to each other, proteins grow into dense fibrous masses that have pathological consequences.

    While the study has some implications for accelerating research efforts for neurodegenerative diseases, it’s more immediate impact will be in biotechnology. Many drugs are proteins, and they are often hampered by unwanted clumping.

    “Protein aggregation is a major headache for pharmaceutical companies,” says Dr. Benedetta Bolognesi, co-corresponding author of the study and Group Leader at the Institute for Bioengineering of Catalonia (IBEC).

    “If a therapeutic protein starts aggregating, manufacturing batches can fail, costing time and money. CANYA can help guide efforts to engineer antibodies and enzymes that are less likely to stick together and reduce expensive setbacks in the process,” she adds.

    Protein clumps are formed using a poorly understood language. Proteins are made of twenty different types of amino acids. Instead of the usual A, C, G, T letters that make up the language of DNA, a protein’s language has twenty different letters, different combinations of which form “words” or “motifs.”

    Researchers have long sought to decipher which combinations of motifs cause clumping and which others enable proteins to fold without error. Artificial intelligence tools that treat amino acids like the alphabet of a mysterious language could help identify the precise words or motifs responsible, but the quality and volume of data about protein aggregation needed to feed models have been historically scant or restricted to very small protein fragments.

    The study addressed this challenge by carrying out large-scale experiments. The authors of the study created over 100,000 completely random protein fragments, each 20 amino acids long, from scratch. The ability for each synthetic fragment to clump was tested in living yeast cells. If a particular fragment triggered clump formation, the yeast cells would grow in a certain way that could be measured by the researchers to determine cause and effect.

    Around one in every five protein fragments (21,936/100,000) caused clumping, while the rest did not. While previous studies might have tracked a handful sequences, the new dataset captures a much bigger catalogue of the different protein variants which can cause amyloid aggregation.

    “We created truly random protein fragments including many versions not found in nature. Evolution has explored only a fraction of all possible protein sequences, while our approach helps us peer into a much bigger galaxy of possibilities, providing lots of data points to help understand more general laws of aggregation behaviour,” explains Dr. Mike Thompson, first author of the study and postdoctoral researcher at the Centre for Genomic Regulation (CRG).

    The vast amount of data generated from the experiments was used to train CANYA. The researchers decided to create it using the principles of “explainable AI,” making its decision-making processes transparent and understandable to humans. This meant sacrificing a little bit of its predictive power, which is usually higher in “black-box” AIs. Despite this, CANYA proved to be around 15% more accurate than existing models.

    Specifically, CANYA is a convolution-attention model, a hybrid tool borrowing from two distinct corners of AI. Convolution models, like those used in image recognition, scan photos for features like an ear or a nose to identify a face, except in this case CANYA skims through the protein chain to find meaningful features like motifs or “words.”

    Attention AI models are used by language translation tools to identify key phrases in a sentence before deciding on the best translation. The researchers incorporated this technique to help CANYA figure out which motifs matter most in the grand scheme of the entire protein.

    Together, these two approaches help CANYA see local motifs up close while also spotting their bigger-picture importance. The researchers could use this information to not just predict which motifs in the protein chain encourage clumping, block it, or something in between, but also understand why.

    For example, CANYA showed that small pockets of water-repelling amino acids are more likely to spark clumping, while some motifs have a bigger impact on clumping if they’re near the start of a protein sequence rather than at the end. The observations align with previous findings researchers have seen under the microscope in known amyloid fibrils.

    But CANYA also found new rules driving protein aggregation. For instance, certain building blocks of proteins, so-called charged amino acids, are normally thought to prevent clumping. But it turns out that in the context of other specific building blocks, they can actually promote clumping.

    In its current form, CANYA primarily explains protein aggregation in yes or no terms, i.e. it works as a so-called “classifier.” The researchers next want to refine the system so it can predict and compare aggregation speeds rather than just aggregation likelihood. This could help predict which protein variants form clumps quickly and which do so more slowly, a vital factor in neurodegenerative diseases where the timing of amyloid formation matters just as much as the fact that it happens at all.

    “There are 1024 quintillion ways of creating a protein fragment that is 20-amino acids long. So far, we’ve trained an AI with just 100,000 fragments. We want to improve it by making more and bigger fragments. This is just the first step but our work shows it is possible to decipher the language of protein aggregation. This is incredibly important for our understanding of human disease but also to guide synthetic biology efforts” concludes Dr. Bolognesi.

    “This project is a great example of how combining large-scale data generation with AI can accelerate research. It’s also a very cost-effective method to generate data,” says ICREA Research Professor Ben Lehner, co-corresponding author and Group Leader at the Centre for Genomic Regulation (CRG) and the Wellcome Sanger Institute.

    “Using DNA synthesis and sequencing we can perform hundreds of thousands of experiments in a single tube, generating the data we need to train AI models. This is an approach we are applying to many difficult problems in biology. The goal is to make biology predictable and programmable,” he adds.

    The study is a joint collaborative effort by ICREA Research Professor Ben Lehner’s lab at the Centre for Genomic Regulation (CRG) and Benedetta Bolognesi’s lab at the Institute for Bioengineering of Catalonia (IBEC). Researchers from Cold Spring Harbor Laboratory (CSHL) and Wellcome Sanger Institute also collaborated in the study. It was funded by “La Caixa” Research Foundation, the European Research Council and the Spanish Ministry of Science and Innovation.

  • University of Tokyo Study Finds Motor Exploration Builds a Stronger Sense of Agency in New Human-Computer Tasks

    Why do people feel they are in control of their own movements, especially when learning an unfamiliar task? New research from the University of Tokyo examines how the brain builds a sense of agency, the feeling that an action and its outcome are self-generated.

    Sense of agency is central to everyday activities, from walking and typing to manipulating objects in the environment. It is also increasingly relevant to modern human-computer interfaces, including virtual reality tools, assistive devices, and emerging brain-machine technologies.

    How the brain detects control

    Researchers often explain agency using the comparator model, where the brain predicts what should happen when a person moves and then compares that prediction with incoming sensory feedback. When prediction and feedback align, the feeling of control tends to strengthen.

    The Tokyo team noted a gap in this framework when applied to learning from scratch, when accurate predictions do not yet exist. In early learning, people often try actions first and only later infer the rules linking movement to outcomes.

    Testing agency during new motor learning

    In the study, participants used a specialized data glove to move a cursor on a screen through finger motions, learning the hand-to-screen mapping by trial and error. At different stages, the researchers assessed how strongly participants felt they controlled the cursor, including when the cursor was subtly shifted in space or delayed in time.

    During the earliest stage, participants relied heavily on timing, judging agency largely by whether the cursor moved in sync with their fingers. With practice, agency increasingly depended on whether the cursor followed the learned mapping, a pattern that was strongest among higher-performing participants.

    Why exploration mattered

    A second experiment reduced motor exploration by having participants imitate presented gestures aimed at reaching target positions. In that setup, the researchers did not observe the same growth in agency, suggesting that imitation alone may not produce the same internal understanding of control.

    The findings point to an important role for active motor exploration in forming a structural representation of how movements cause outcomes. The researchers argue this kind of rule discovery can help build more robust agency, with potential implications for rehabilitation training, VR interaction design, and future interface development.

  • Long-term gorilla study finds friendship can boost survival yet raise health risks

    Friendship among mountain gorillas can pay off in some ways while creating new risks in others, according to a new long-running field study tracking health and reproduction. Researchers say the mixed outcomes may help explain why some individuals are more sociable than others.

    The analysis drew on more than 20 years of observations of 164 wild mountain gorillas in Volcanoes National Park, Rwanda. Scientists compared patterns of close social bonds with measures including illness, injuries from conflict and reproductive success.

    Benefits shift with group size

    The study found that the advantages of social ties depended heavily on a gorilla’s group environment, including group size and stability. What looked beneficial in one setting could become costly in another as competition and exposure to disease changed.

    For females, strong and stable relationships were generally linked to less illness, but the pattern was not uniform across groups. In smaller groups, more socially connected females tended to fall ill less often but also had fewer offspring.

    In larger groups, the relationship appeared to flip in important ways, with friendly females showing higher birth rates while also experiencing illness more frequently. Researchers suggested that crowded social settings may increase exposure to pathogens even as they improve access to support and mating opportunities.

    Male bonds trade illness for safety

    For males, tight social bonds were associated with getting ill more often, but those connections also came with a protective effect. Well-bonded males were less likely to be injured in fights, a major threat in a species where conflict can be severe.

    Lead author Robin Morrison of the University of Zurich said the findings suggest it is not simply a case of more social contact causing more disease. One possibility raised by the team is that maintaining close ties may carry energetic and stress-related costs for males, potentially affecting immune function.

    The work, conducted with the Dian Fossey Gorilla Fund and researchers at the Universities of Exeter and Zurich, underscores how social behaviour can be shaped by competing pressures. The authors argue that there may be no single best social strategy, because the optimal level of sociability shifts with sex, group context and life stage.

    The paper was published in the Proceedings of the National Academy of Sciences and adds to broader evidence that social environments can strongly influence health and lifespan across social mammals, including humans. At the same time, it cautions against assuming that more friends is always better, even in highly social species.

  • African Starlings Show Long-Term Reciprocity, Offering New Evidence of Friendship-Like Bonds in Birds

    Researchers studying African starlings say the birds build durable social ties that can resemble human friendship, repeatedly helping specific partners over time. The findings add rare long-term evidence that cooperation among non-relatives can be sustained through reciprocity.

    The study, led by Alexis Earl with colleagues in Professor Dustin Rubenstein’s research group, drew on observations and genetic sampling spanning roughly two decades. By pairing behavioral records with DNA data, the team could separate assistance to relatives from help offered to unrelated group members.

    What the long data revealed

    Biologists have long understood why animals support close kin, since assisting relatives can boost shared genetic success. What has been harder to demonstrate is whether animals reliably invest in non-relatives with an expectation of future return.

    Across many breeding seasons on the East African savannah, the researchers documented thousands of interactions among hundreds of birds living in complex groups. They found that while helpers often favored relatives, many also provided consistent support to particular non-relatives even when kin were available.

    That pattern points to reciprocity rather than simple proximity or chance, the authors argue, because the same unrelated partners kept reappearing in helping roles over multiple years. The work suggests some starling relationships remain stable enough to function as long-term cooperative partnerships.

    Why reciprocity is hard to prove

    Detecting reciprocal helping typically requires large datasets collected over long periods, since favors may be returned much later. Shorter studies can miss these delayed exchanges and may misclassify them as one-off acts.

    The authors say the next step is to examine how these bonds form, what keeps them resilient, and why some cooperative pairs break down. More broadly, the results support the idea that similar reciprocity may exist in other animal societies but remains under-detected due to limited multi-year tracking.

  • Study links parental death to higher bullying risk in teens, with rural girls most affected

    Losing a parent or primary caregiver is one of the most disruptive events a child can face, and new research suggests it may also change how young people are treated at school. A large survey-based study found higher odds of bullying victimization among youth who experienced parental death.

    The study, led by researchers at Boston University School of Public Health and collaborators in China, analyzed responses from about 21 000 children aged 10 to 17. It was published in the Journal of Affective Disorders and drew on data collected between 2019 and 2021 in southwestern China.

    Who faced the biggest rise?

    Researchers reported that the link between bereavement and bullying varied by age, sex, the parent who died, and where children lived. Adolescents aged 13 to 17, girls, and students in rural areas showed the highest increases in reported bullying after a parent’s death.

    Maternal death was associated with a higher bullying risk specifically among boys in the sample, a pattern the authors said may reflect a uniquely protective role of maternal support for sons. The study emphasizes that family disruption can affect social behavior, confidence, and peer relationships during sensitive developmental periods.

    What the survey data showed

    In the study group, nearly 3 percent of participants reported experiencing a parental death. More than 15 percent said they were being bullied at school, highlighting how common peer victimization remains across adolescence.

    While most parental deaths captured in the dataset occurred before COVID-19 became widespread, global estimates indicate the pandemic left millions of children bereaved. Separate international analyses have estimated that more than 8 000 000 children worldwide lost a parent or primary caregiver due to pandemic-related causes.

    Support that lasts beyond the funeral

    The researchers argue that support for bereaved children should be long-term and tailored, combining emotional care with help navigating peer dynamics. They pointed to counseling, engagement of remaining caregivers or extended family, and school programs adjusted to developmental stage and cultural context.

    The study also calls for schools to train staff to recognize grief-related vulnerability and to strengthen overall school climate. Creating an inclusive environment, the authors suggested, may reduce bullying risk and help bereaved students regain stability as their needs evolve.

  • Chimpanzee call combinations hint at language roots: What a major Taï Forest study found

    New research based on years of field recordings suggests wild chimpanzees can combine calls in ways that shift or expand meaning, a trait long seen as central to human language. The findings challenge the idea that complex, rule-like vocal communication is uniquely human.

    Scientists studied thousands of vocalisations from three chimpanzee groups in Taï National Park in Côte d’Ivoire. They analysed how the meanings of 12 call types changed when paired into 16 different two-call combinations across many everyday situations.

    How chimpanzees change call meaning

    The researchers report four main patterns that alter meaning when calls are combined, including compositional pairings that add information. Other combinations appear to clarify context, helping narrow a call’s meaning depending on what follows.

    The study also describes non-compositional, idiom-like combinations, where two familiar calls together convey a meaning that is not a simple sum of the parts. That kind of structure resembles a key feature of human language, where fixed phrases can carry distinct meanings.

    Why the context matters

    Previous work on primate call combinations has often focused on narrow scenarios such as predator alerts, with only a small number of known combinations. In Taï, the team found a broader and more flexible set of pairings used across feeding, travelling, social bonding and other contexts.

    Because chimpanzees and bonobos are humans’ closest living relatives, the results feed into a larger debate about language evolution. The authors argue these combinatorial abilities may have deeper evolutionary roots than once assumed, though they also note more comparative research is needed.

    What this means for language origins

    The findings do not mean chimpanzees use human-like grammar, but they do suggest vocal communication can be more generative in great apes than previously documented. That could shift how scientists frame the early building blocks that eventually supported human language.

    Researchers also warn that long-term fieldwork is becoming harder as human pressures grow on wild chimpanzee populations. Continued observation in natural habitats, they argue, is essential to map the full range of great ape communication.