This week in psychology: good conversations may depend on not agreeing quite as much as we think, different forms of exercise may support different cognitive skills in children with ADHD, and two parts of the brain can apparently possess similar information while doing very different things with it.
Three studies worth knowing about, and rather conveniently, all three have something to say about a recurring psychological mistake: assuming that if we know what the ingredients are, we must understand the recipe.
Good Conversations May Need a Little Disagreement
Most advice about conversation tends to emphasise common ground. Find something you both like, demonstrate that you understand the other person, nod sympathetically at appropriate intervals, and perhaps resist checking your phone while they are halfway through describing their divorce.
There is good psychological reason for some of this. Similarity and interpersonal synchrony can make interactions easier, while feeling understood is generally preferable to the conversational experience of explaining yourself to somebody who appears to be waiting for their turn to speak.
But complete agreement is not necessarily what makes a conversation interesting.
O’Nell, Sanchez and Finn (2026) examined 1,656 conversations between strangers, tracking how semantically similar each person's contributions were as the conversations unfolded. Rather than asking simply whether successful conversational partners talked about similar things, they looked at how that similarity changed over time.
The resulting pattern was rather elegant.
Enjoyable conversations tended neither to remain closely aligned nor wander randomly from topic to topic. Instead, conversational similarity fluctuated across several timescales in a pattern resembling what mathematicians call pink noise. People aligned, drifted apart, explored something slightly different, and then found their way back.
Pink noise sounds more mysterious than it needs to. The important idea is that change occurs at several different scales rather than being either perfectly regular or completely random. Similar patterns appear in systems ranging from heartbeats to music, and apparently now in two strangers trying not to make small talk unbearable.
The researchers found that conversations showing more of this structure were associated with greater enjoyment and interpersonal connection. Evidence for similar patterns also appeared in additional conversational datasets (O’Nell et al., 2026).
There is something intuitively satisfying about that result. A good conversation often does involve agreement, but agreement provides somewhere to start rather than somewhere to remain. One person introduces an idea, the other modifies it, challenges part of it, connects it to something unexpected, and eventually both arrive somewhere neither quite began.
That process is different from conflict. The study does not suggest that periodically disagreeing with your friends will make them enjoy your company more. The observed relationships were modest, with correlations around .11 for enjoyment and .09 for connection in the largest dataset, and the research cannot establish that producing this pattern causes conversations to improve.
Still, it provides a useful way of thinking about conversational chemistry.
Perhaps enjoyable conversation depends partly on balancing two opposing needs: enough similarity that we feel understood, but enough difference that there remains something worth discovering.
Someone who disagrees with everything is exhausting.
Someone who agrees with everything may eventually become furniture.
Exercise and ADHD: The Activity May Matter
“Exercise helps ADHD” has the pleasing simplicity of advice that fits comfortably on an infographic.
The evidence is rather more interesting.
A new systematic review and network meta-analysis published in Molecular Psychiatry compared different forms of exercise across 40 randomised controlled trials involving 1,563 children and adolescents with ADHD (Hou et al., 2026). Rather than treating cognition as a single outcome, the researchers examined three components of executive functioning: working memory, inhibitory control and cognitive flexibility.
These overlap, but they are not interchangeable.
Working memory allows us to temporarily hold and manipulate information, such as remembering the beginning of a sentence while reaching the end. Inhibitory control helps us suppress an immediate response when another action would be more appropriate. Cognitive flexibility allows us to shift between rules, perspectives or strategies when circumstances change.
The analysis suggested that different activities may benefit these abilities differently. Mind-body exercise ranked particularly well for working memory, while combat sports ranked highest for inhibitory control and cognitive flexibility (Hou et al., 2026).
That immediately raises an interesting possibility.
Exercise may not influence executive function simply because hearts beat faster and bodies move around. Different activities also make different cognitive demands.
A martial art, for example, requires somebody to monitor an opponent, inhibit premature responses, select between possible actions and rapidly adjust when circumstances change. Mind-body activities may place greater demands on sustained attention, sequencing and coordinated control.
If those cognitive demands matter, then asking whether “exercise” helps cognition may be a little like asking whether “reading” teaches somebody something. It probably matters what they are reading.
The researchers also found evidence of nonlinear relationships between exercise dose and cognitive outcomes, suggesting that more was not necessarily better indefinitely. That is important because interventions have an unfortunate tendency to acquire wonderfully precise recommendations before the underlying evidence has earned them.
There are reasons for caution here. Network meta-analysis allows researchers to compare interventions even when many have never been tested directly against one another. Being statistically ranked first therefore does not mean combat sports have comprehensively defeated every other activity in some grand tournament of ADHD treatments.
The individual trials also varied in their programmes, participants and quality. The apparent differences between exercise types should therefore be treated as promising evidence about where researchers ought to look next, not prescriptions for exactly how many minutes of taekwondo a child requires.
And exercise remains an adjunct, not a replacement for established ADHD care.
Still, the study complicates the usual advice in a productive way. ADHD is not simply “difficulty paying attention”, executive function is not one mental muscle, and exercise is not one intervention.
Once those distinctions are made, the more useful question becomes not simply whether physical activity helps, but what kinds of activity might help which cognitive processes, for whom, and why.
That is a considerably less convenient slogan.
It is also a better research question.
Knowing Something Is Not the Same as Using It
Brain imaging has given psychology an extraordinary ability to identify information represented in different parts of the brain.
It has also created a very understandable temptation.
If a brain region contains information about something, perhaps that tells us what the region is for.
New research in Nature Neuroscience illustrates why the relationship is not quite that tidy.
Ruff and colleagues (2026) studied two rhesus monkeys performing a decision-making task involving both visual motion and potential rewards. The researchers recorded activity from a brain region specialised in processing visual motion and from the dorsolateral prefrontal cortex, which has broader involvement in cognition and decision-making.
Both areas contained information about motion and reward.
At first glance, that creates a puzzle. If both regions know broadly the same things, why does the brain need both?
The answer appears to lie in how that information is organised.
In the visual region, representations of motion and reward remained more separable. The prefrontal cortex combined the information in a way more closely related to the decision the animal eventually made (Ruff et al., 2026).
The distinction became more convincing when the researchers moved beyond simply observing neural activity. Computational modelling predicted that stimulating the two regions should affect behaviour differently, and causal stimulation produced distinct behavioural effects broadly consistent with those predictions.
So two regions can represent similar ingredients while performing different computations with them.
That sounds technical, but the conceptual point is straightforward.
Imagine two employees receiving the same spreadsheet. One organises the figures into categories. The other combines them into a recommendation about what the company should actually do. Looking only at which information appears on each screen would miss the difference between their jobs.
Neuroscience can face something similar.
Finding that a region contains information about reward, faces, fear or anything else does not necessarily tell us what contribution that information makes to behaviour. The relevant question may be not simply what does this region represent?, but what transformation does it perform on what it represents?
That distinction matters well beyond this particular experiment. Popular neuroscience often treats activation as explanation. A region becomes active during an experience, a coloured blob appears on an image, and we are only a headline away from announcing that scientists have found where jealousy lives.
The brain is unlikely to be quite so accommodating.
There are important limits to the current study. It involved two rhesus monkeys performing a highly controlled perceptual decision task, so we should resist immediately applying the computational distinction to every part of human cognition. Two monkeys have carried enough theoretical responsibility already.
But the causal component makes the basic result particularly interesting.
The same information can exist in multiple places without those places doing the same thing.
Sometimes the important question is not what the brain knows.
It is what the brain does with what it knows.
Simply Put
These three studies look unrelated until we consider what each one is actually measuring.
A conversation cannot be understood simply by counting agreement. Exercise cannot be treated as one uniform input producing one uniform cognitive benefit. Brain activity cannot be interpreted merely by identifying where particular information appears.
In each case, organisation matters.
The pattern through which two people move between similarity and difference matters. The cognitive demands embedded within an activity matter. The way a neural system transforms information matters.
Psychology has traditionally been very good at identifying variables: attention, similarity, exercise, reward, inhibition, brain activity. The harder task is understanding what happens between them.
That tends to be where the explanations become less neat.
Fortunately, it is also where they usually become more interesting.
References
Hou, Y., et al. (2026). Comparative effects of different exercise interventions on executive function in children and adolescents with attention-deficit/hyperactivity disorder: A systematic review and network meta-analysis. Molecular Psychiatry. https://doi.org/10.1038/s41380-026-03943-3
O’Nell, K. A., Sanchez, J. S., & Finn, E. S. (2026). Pink noise in semantic similarity predicts enjoyment and connection in conversation. Science Advances. https://doi.org/10.1126/sciadv.aeb2684
Ruff, D. A., et al. (2026). Distinct neural geometries support different causal contributions to decision-making. Nature Neuroscience. https://doi.org/10.1038/s41593-026-02514-7
Table of Contents
