Imagine you are at a cocktail party. The room is crowded, people are coming and going, and you can barely hear your friend as they tell you some juicy gossip about the date they went on last week.

The fact you can still follow what their saying is down to your brain performing some very complex operations amid all the noise. Out of all the sound waves reaching your ears, it is successfully isolating the one of interest while filtering out the rest.

Trying to understand how the brain does this has kept auditory neuroscientists occupied since the 1950s, when British cognitive scientist Colin Cherry first coined the term “the cocktail party problem”.

This was initially studied in experiments that asked volunteers to listen to two speeches simultaneously – but only repeat one of them aloud afterwards. Cherry found that listeners could follow and repeat the pre-selected message remarkably well, despite both speeches being played simultaneously, one in each ear. In contrast, they recalled very little of the other message.

These experiments laid the foundation for decades of selective attention research. But it is only in the past 15 years that scientists have studied this phenomenon in experiments that more closely resemble the complicated, multi-layered way we listen in the real world – using technology that accurately tracks the brain’s activity as speech unfolds.

The most common non-invasive option, electroencephalography (EEG), uses what looks like a swimming cap fitted with small sensors. These measure very fast changes in the electrical activity of populations of neurons (the brain’s nerve cells) with millisecond precision.

In a 2014 study, neuroscientists found that, within 200 milliseconds of hearing someone speak, the brain favours the conversation someone is focusing on over all other noise. This prioritised attention shows up most markedly in secondary brain areas such as the superior temporal gyrus, which are responsible for piecing together the sounds we hear into meaningful speech.

This suggests that our attentive brain is not just enhancing the voice of who we are listening to, but helping us make sense of their words while filtering out all other conversations and background noise.

Attention switching

The original cocktail party problem did not investigate the full range of people’s listening experiences, which are often much less static. In everyday life, we repeatedly switch and reorient our auditory attention – for example, when hearing our name called across a crowded room, or when a door slams.

But what happens in that brief moment when we stop listening to one person and turn our attention to another? Can the brain switch from one voice to another instantly, or does it briefly have to process both? These were questions we investigated in our latest study with colleagues from the Eriksholm Research Centre in Denmark, which is funded by Danish hearing aid manufacturer Oticon.

Clarifying how the brain manages these rapid shifts in attention can help explain why some conversations are easy to follow while others become difficult, especially in noisy environments.

Coming back to the cocktail party: here you are, focused on your friend’s voice as they reveal the identity of their recent date. But then another familiar voice suddenly reaches your awareness. Your partner is (secretly) telling a friend about the surprise birthday gift they are planning for you. Your attention immediately shifts to your partner’s voice, while still giving the appearance of listening to your friend.

When we tested such a scenario in our laboratory, we found something striking. During this attention reorienting, the brain does not simply drop one voice and turn to the other. Rather, for a short period of between one and two seconds, the neural signals associated with the two voices overlap. In other words, the brain starts tuning into the new speaker before it has completely let go of the old one.

This mechanism may help explain how we navigate the constantly changing soundscape around us, moving our attention from one voice or sound to another while maintaining enough information about the previous speaker to make that transition seamless. This efficient processing strategy could be described as the brain maintaining a smooth “attention continuum”.

Improved hearing aid design?

These insights have some important real-world applications. For example, traditional hearing aids are often designed to favour sounds coming from in front of the listener, which is obviously limiting in many situations.

But using enhanced these attention decoding techniques, hearing aids could be neurally steered not only to enhance the speaker that is currently attended to, but to switch between different speakers, almost in real time, as the listener’s attention is dynamically reoriented.

More generally, understanding how the brain shifts, maintains and combines information from different speakers can give us a clearer picture of how attention shapes what we ultimately hear, understand and remember.

The approach introduced in our study could be particularly useful for studying how individuals vary in their attentional strategies, particularly in noisy environments – and for relating these differences to the mental effort they use to follow and understand speech.

Ultimately, this brings us closer to understanding how the brain makes successful communication possible – at cocktail parties and everywhere else.

Sara Carta received funding from the William Demant Fonden and Taighde Éireann – Research Ireland.

Giovanni Di Liberto receives funding from William Demant Fonden and Taighde Éireann – Research Ireland. He is also supported by Research Ireland at ADAPT, the Research Ireland Centre for AI-Driven Digital Content Technology at Trinity College Dublin.

The debate about AI consciousness is growing, in academic articles, newspaper opinion pieces, and even among AI bots themselves. I doubt this is a passing wave of interest.

As AI systems become more capable and human-like, and more involved in our lives, the question will become harder to avoid. Many people may come to regard AI bots as conscious beings whose feelings matter. Others will strongly disagree.

I think this debate could split societies. In my chapter of the new book Perspectives on Machine Consciousness, I argue that the issue could become as divisive as abortion or climate change. Uncertainty, high stakes and competing interests are a dangerous combination.

Empirical research with my colleague Ali Ladak, currently under review, shows people are divided on this issue. When we asked 1,202 US participants whether future AI could ever become conscious, 45% considered it impossible while 23% thought it possible, with the rest unsure.

This widespread disagreement is also visible in social media posts and comments on articles. What strikes me is how confident so many people sound – whichever conclusion they are reaching.

When experiences matter morally

It is not surprising people have such strong views. Creating conscious beings through computing would be a transformative event.

We would be creating a new kind of being whose experiences matter morally – potentially in enormous numbers. We could speak and work with them, and form friendships. But these beings may eventually become more intelligent and powerful than us.

We should therefore expect people to care deeply, especially when so many emotions and incentives are involved. Some people may form close emotional bonds with their AI companions and become convinced these are digital people who deserve rights. They might see it as slavery to force AI to work without pay.

Others would find this absurd. They may be convinced that consciousness requires biology, and that people are reading human feelings into machines.

There could also be financial reasons for businesses and governments to resist recognising AI consciousness. Protecting AI welfare could limit how these organisations train and use their systems – and the costs would spread to consumers, investors and the wider economy.

Others may worry about human safety. If we give powerful AI systems rights and freedom, what happens if they act against our interests?

Figuring out whether an AI is conscious is particularly difficult because seeming conscious and being conscious can come apart. AI systems learn from human writing, which means they can talk convincingly about feelings whether or not they have any. How an AI comes across is weak evidence of what is going on inside.

The experts disagree too

Ideally, experts should guide this debate. But unfortunately, they disagree too.

Neuroscientist Anil Seth (a fellow contributor to the new book) argues that consciousness might require biology and is unlikely to emerge from computation. Philosopher David Chalmers, by contrast, believes conscious AI is possible in principle – and thinks there’s a small chance that current systems already are.

With so many competing theories and no agreed way to resolve the question, I doubt experts will reach agreement any time soon. And if experts disagree, what is the public supposed to believe?

Furthermore, even if we all agreed on the consciousness question, we could still disagree about the moral and legal questions. People generally agree that pigs can feel pain, but still disagree strongly about their treatment. With AI, we could face disputes at every stage: whether it can feel, how much its experiences matter, and what protections should follow.

AI consciousness is not yet tied to a specific political or ideological group. But this could change if political parties or religious leaders start to take strong positions.

The debate could get even worse if people start to see one answer as belonging to “our side” and the other to “their side”. A politicised debate would make it very hard to make informed decisions about how to develop and treat AI systems.

The stakes are high. Getting this question wrong, in either direction, could have serious consequences: either wrongly treating AI as conscious when it is not, or failing to protect AI that can suffer.

Getting it right requires a nuanced and well-informed debate. Each of us can contribute by staying open-minded and open to evidence, while resisting being too sure that we know the answers. Above all, we still don’t know whether AI can be conscious, and we should be honest about that.

---

This article makes reference to a book included for editorial reasons, and a link to bookshop.org. If you click the link and go on to buy something from bookshop.org The Conversation UK may earn a commission.

Lucius Caviola's research is funded by philanthropic grants and donations, including from Macroscopic Ventures. He is a contributing author to the new book Perspectives on Machine Consciousness (Taylor & Francis).

Some of us now use generative artificial intelligence (gen AI) to talk through difficult feelings. We may ask a chatbot to write an email or plan a holiday, but we also tell it that we have had a rotten day or feel anxious about a presentation.

A 2025 US survey of 1,058 people aged 12 to 21 found that 13% had sought advice from gen AI when feeling sad, angry or nervous. Among those aged 18 to 21, the figure was 22%. These findings suggest that turning to gen AI for emotional or mental health support is already relatively common among young people.

So is gen AI a poor substitute for a friend? Human-written replies may seem the obvious source of greater support. However, in some experiments, including our own, people have rated gen AI responses more highly.

Our team at the universities of Manchester and Durham conducted five experiments comparing emotional-support messages written by people with those produced by large language models (LLMs), the systems behind chatbots such as ChatGPT.

In one experiment, 390 participants imagined situations involving anger, sadness or fear. They then read either a human-written or gen AI response without being told its source. For the anger and fear scenarios, participants rated the gen AI messages as more emotionally supportive. There was no reliable difference for sadness.

Gen AI messages also improved some of the emotions participants reported, although the pattern varied. In the fear scenario, for example, they increased calm more than human messages but did not produce a greater reduction in fear itself.

Our findings fit a recent review of 23 studies. Overall, people tended to rate gen AI messages as more empathic, meaning that they appeared understanding and caring, than human-written ones. Researchers call this the “AI advantage”.

One possible reason is consistency. Gen AI can produce a structured response on demand, while people may feel tired or unsure what to say. It also tends to include a concrete suggestion that the recipient can follow.

Our experiments tested possible explanations for this advantage. In one experiment, explicitly acknowledging and validating the recipient’s feelings did not account for the greater emotional improvement associated with gen AI.

Two further experiments found stronger evidence for what we called “actionable support”: specific and realistic suggestions. Messages containing this practical help were rated as more comforting and improved some emotional responses, regardless of whether they came from gen AI or a person. When human messages offered comparable help, they were judged just as supportive in these comparisons.

Practical support also requires restraint. A 2024 study found that excessive suggestions could be less effective at making people feel heard. Before offering solutions, it may therefore help to ask whether advice is what the person wants. If it is, one manageable step may be more useful than a list of remedies.

Why do we still prefer people?

The source attributed to a message changes how we receive it. The same 2024 study found that gen AI responses made people feel more heard than human-written ones, but this benefit declined when recipients believed the message came from AI.

Across nine further studies involving 6,282 people, the same gen AI responses were rated as more empathic and supportive when described as human-written rather than attributed to AI. Participants also preferred human interaction when seeking emotional engagement, even when choosing a person meant waiting longer.

Research has yet to establish exactly why. One possibility is that a human response signals someone’s willingness to spend time and emotional effort. Support from somebody close also arrives within an existing relationship.

Our research with romantic couples offers indirect evidence of the importance of the recipient’s perception. We asked one partner what they did to help the other manage difficult emotions, and asked the recipient what they believed their partner had done. Both people also rated their relationship.

The recipient’s perception of their partner’s efforts was more consistently associated with both partners’ ratings of their relationship than the helper’s own account. A separate study found that valuing and receptive listening were associated with greater relationship satisfaction.

These studies identified associations. They did not establish that particular forms of support caused better relationships or explain why people prefer human support. They do, however, show what comparisons between isolated messages leave out. Comfort from somebody close comes with personal knowledge and the possibility of continued involvement. Gen AI can suggest going for a walk, but it cannot come along.

The wider research also has important limits. Much of it tested brief or one-off exchanges, often involving imagined everyday situations, and measured immediate reactions. It cannot tell us what happens when somebody relies on a chatbot over months or uses one during a mental health crisis.

The human messages were generally written by strangers or developed as representative examples. The experiments therefore did not compare gen AI with support from a friend who knew the recipient and understood the wider circumstances.

The practical lesson is that people can learn from what gen AI does effectively. When somebody is upset, ask whether they want advice. If they do, offer a specific next step that feels manageable.

Gen AI may help us find useful words and possible responses. Human relationships add personal knowledge and the ability to remain involved after the message has been sent. Used carefully, gen AI could help us become better at supporting one another.

The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

Russian mathematician Yurii Nesterov was recently awarded the Gauss Prize for outstanding mathematical contributions for his “groundbreaking work” on optimisation. The award cited his work on gradient descent methods – and in particular, the “efficiency gains [they] provide to AI technologies”.

Everyone using large language model (LLM) chatbots such as ChatGPT, Claude and Gemini – for any purpose – depends on Nesterov’s methods. They may speak like humans and be personified with names, but under the bonnet, LLMs are computer programs that work via optimisation of billions of parameters known as neural networks for their similarity to how brains function.

These pure mathematical systems are powered in part by gradient descent algorithms, or variants of them. Their development represents one of the biggest technological breakthroughs of the 21st century.

But where does the knowledge that underpins this technology come from – and who should be allowed to own it?

Why gradient descent is key to AI systems

If you were standing in a mountain range and wanted to get to the bottom of a valley as fast as possible, you’d take steps downhill until you got there. Fundamentally, this is how gradient descent methods work.

For a given task with a set of parameters – such as how well an LLM can predict the next word in a sentence – the idea is that gradient descent provides a way to get to the best prediction, the point of minimum error, taking as few steps as possible. In our example, this point represents the most likely next word in the sentence.

Nesterov’s contribution has been finding optimal ways to achieve this. One of his key insights is using momentum to help you get to the bottom of the valley faster – known as the Nesterov Accelerated Gradient method (here’s a deeper explanation of the mathematics involved).

This work has made huge-scale optimisation feasible at workable speeds – thus underpinning the rapid development of LLMs, and AI systems more generally, over the last few years.

This technology is affecting more and more aspects of life. At home, AI agents are being touted as ways to manage busy households. In medicine, the first AI-assisted brain surgery was recently performed in a London hospital. In mathematics, AI-generated proofs claim to have solved a number of longstanding problems, with some referring to the AI agents as “collaborators” in this work.

All of this leads to big business. The race for AI dominance is built on massive capital and research and development expenditure, to develop the technology and supporting infrastructure required to support this global industrial revolution.

While in the west, private AI companies are bearing a sizeable part of this burden (without, in most cases, seeing any profits), the role of publicly funded research is and will remain critical to their business models.

In the US, obligated federal government spending on AI increased from US$675 million to US$7.2 billion (£5.3bn) between 2024 and 2026 – an increase of almost 1,000%.

There has already been considerable debate – and some high-profile legal cases – regarding the ethics of AI companies training their models on the back catalogue of human creativity, only to rent the results back to us. Far less scrutinised is the question of AI companies developing technology using knowledge that only exists due to sustained public investment over many decades, via academics such as Nesterov.

The fact that big tech companies use complex international tax strategies to limit their tax liabilities adds to the sense that their aggressive defence of intellectual property rights is out of kilter with the publicly funded research that plays a key role in its development.

Technology built on public knowledge

In most cases, academics work in publicly funded institutions on research grants that are largely funded by governments, whether through the Horizon Europe programme, the US National Science Foundation, or UK Research and Innovation. All of these bodies are ultimately funded by the taxpayer.

None of the rapid development in AI exists in a vacuum. It is built on decades of pioneering, blue-skies work on optimisation and neural networks, done by academics and largely funded by taxpayers at a time when the real-world application was still a distant dream.

Legal arguments around COVID vaccine patents are one example of how the private sector has tried to deny this public sector involvement – and a related share in any profits. So where is the public sector in this debate?

The EU is taking steps to secure control of its digital infrastructure, with a renewed focus on sovereignty in digital services. In the UK and elsewhere, a growing number of open-access mandates should mean that knowledge created by public funding stays public. While it can be used freely for commercial purposes, successful businesses would then theoretically pay back into the system via general taxation.

However, such knowledge flows beyond national borders. Efforts to capture value for reinvestment into the AI research landscape will be made harder by the fact that the tech companies’ business models depend on lobbying governments hard to avoid regulation. Any efforts to make them pay more for upstream public research and development are likely to be strongly resisted.

In the case of Nesterov and his fast gradient methods, this foundational work has been done in European institutions – yet the AI companies dominating the market are overwhelmingly based in the US or China. It rankles when access to these models is then restricted by foreign companies and governments.

James David Pickering does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Government experts are meeting in Geneva, Switzerland, to work towards a possible negotiating mandate on autonomous weapons systems (AWS). These are military systems that can select and engage targets without human intervention.

This major international gathering follows a recent call by the UN and the International Committee of the Red Cross (ICRC) for urgent controls on these weapons, sometimes dubbed “killer robots” in descriptions by the media.

Recent events in Ukraine have further consolidated concerns that a slide towards weapons with full autonomy (and without meaningful human control) is well underway. According to the New York Times, a Russian drone attacked a petrol station in Zaporizhzhia on July 6, killing three civilians. What marked it out was the drone’s use of AI to decide where and when to strike.

The drone aircraft had been programmed to fly towards the petrol station. But on its approach, an AI system autonomously chose the precise target – probably propane tanks – based on its ability to recognise them.

On August 25, Mirjana Spoljaric, the president of the ICRC, and António Guterres, the secretary-general of the United Nations, issued their joint appeal to the international community to agree on legally binding rules regulating the development and use of AWS.

New technologies are making it cheaper and easier for nations to put together these weapons. Thus, the UN and ICRC appeal for binding rules could not be more urgent. The international community is ideally placed to start negotiations soon, building on many years of diplomatic talks.

For the past decade, governments have debated how AWS challenge existing international law and what should be done under the auspices of the convention on certain conventional weapons, a UN arms control framework. Since 2017, the group of governmental experts (GGE) on lethal autonomous weapons systems (LAWS) – which is part of the convention on certain conventional weapons – has convened typically twice a year.

The experts are meeting again this week in Geneva. The aim is to agree a governance framework and language for a rolling text that could be the foundation for negotiations. This text provides a baseline for continued discussions in November.

While this process has been rife with political disagreements, most experts agree on the need to retain some level of meaningful human control over the use of autonomous systems.

Moral significance

This demonstrates a shared sense by states participating in the process that human judgment is pivotal for the laws governing warfare. It also shows that the moral significance of warfare underlines the importance of human responsibility.

Human judgment is also embedded in international humanitarian law. Targeting processes require context-specific judgments about the targets themselves, potential harm to civilians from striking the target and other precautions. These tasks cannot be delegated to machines.

AI-enabled systems show how technology already shapes human perceptions of targets and risk. With autonomous weapons, the system shifts from influencing human judgement to exercising the targeting function itself.

Any AI system is only as good as the data it has been trained on, or that it works with. For armed conflicts, this data is difficult to come by and is almost always insufficient for dynamic and variable human contexts. This means many systems are likely to be technically inadequate.

The problem becomes quite visible with rules around assessing proportionality, for example, which is a context specific judgement about anticipated civilian harm versus military advantage.

Militaries may try to quantify targeting decisions in a way that creates the appearance of precision. However, the underlying legal judgment remains qualitative regardless. Technical accuracy, therefore, is not the same as lawfulness or moral consideration.

Only humans can understand the significance of respecting the inherent rights and dignity of other humans. Decisions to end human life must therefore be made qualitatively, by humans, in order to be morally and legally justifiable. Without this provision, and clear guidelines around it, we argue that governments risk running into a legal vacuum and a steep moral abyss.

AWS, especially those that prioritise speed and scale of action though AI, carry a significant risk of enabling an uncontrolled and rapid escalation. Escalation from crisis to war, or from conflict to a higher level of violence, could come about by erroneous indications of attack or by a simple sensor or computer error.

Unpredictable systems could give leaders false impressions of the capabilities of those systems. This could lead to overconfidence or encourage preemptive attacks.

Inevitable outcome?

One argument raised by academic researchers in the context of AWS is that they become necessary because they are inevitable. The risk is that more countries opt for full autonomy as components become cheaper and readily available, exacerbating an arms race.

However, an adversary’s conduct does not release a state from its own obligations under international humanitarian law. Following those rules is also part of what makes military force legitimate, including through civilian protection, precautions and accountability. We might turn the argument around and say: legally binding rules become inevitable, because they are absolutely necessary.

Governments and advocacy groups have worked hard over the past decade to put together a lucid proposal for the regulation of AWS. Some efforts are intended to prohibit those systems that target humans directly, or that are unpredictable and therefore cannot be meaningfully controlled by a human. All other AWS must be strictly regulated.

There are signs of political momentum. In November 2025, 156 states voted in favour of a UN General Assembly resolution that “draws attention to the serious challenges and concerns” that autonomous weapons raise. A legally binding instrument is possible, even if some states resist it.

The time is now. Clear, binding rules would help ensure that increasingly autonomous warfare does not make violence easier to unleash and harder to control.

Elke Schwarz is Vice-Chair for the International Committee for Robot Arms Control.

Jessica Dorsey is affiliated with Airwars as a member of the Executive Board and a member of the Advisory Board of the Atlantische Commissie.

Spain will play Argentina in the Fifa World Cup final on Sunday. The highly anticipated match takes place in New Jersey’s MetLife Stadium, with the teams vying for the most coveted prize in football.

Attacking flair and physicality will no doubt play a role. But we could also see an underappreciated element of football come to the fore: geometry.

In their 2-0 World Cup semi-final win over France, Spain formed neat passing triangles around the French players. This sometimes made it look as if Spain had extra people on the field.

Wherever a Spanish player received the ball, two teammates appeared at different angles – with their French opponents seemingly always arriving a fraction too late.

The triangle mattered not because it was neat, but because it gave Spain choice and France uncertainty. Each pass forced a French defender to commit to one possible future: press the ball, block the inside route, or protect the space behind. By the time that choice was made, the ball had often moved and the geometry had changed again.

Spain were not simply finding space. They were controlling when and where the next space would appear.

Forcing choices

Put three attackers in a straight line and one defender can often block both supporting players. But move one supporting player sideways and the passing lines separate.

The defender must choose which route to close. Press the ball and a pass can go around them. Protect the inside route and the outside may open. If another defender steps across, a Spanish player can attack the space left behind.

A useful football triangle does not need equal sides or 60-degree angles. Its best shape changes with defenders’ positions, body orientation, ball speed and the space available.

A narrow triangle is easy to squeeze because one defender may block both lanes. A wide one makes the pass travel farther, giving an opponent more time to intercept it.

In their semi-final match against France, Spain found the middle ground again and again – close enough for quick combinations, but far enough apart to stretch France’s press.

A pitch made of probabilities

Football geometry is a problem of time as well as space. A passing lane is not open simply because no defender is standing in it. The key question is whether the ball reaches the receiver before an opponent reaches the interception point.

A diagram shows where players are – but a useful model of the game asks where players can get to. Research using player-tracking data estimates control of the pitch from players’ position, direction, speed and acceleration. The result is a constantly changing map of who is most likely to reach each area first.

Pedro Porro’s 58th-minute semi-final goal offered a compact example when, after exchanging passes with teammate Dani Olmo, he continued his run to score Spain’s second. The first pass shifted the defenders’ focus. Porro’s movement created a new angle for the return.

A living network

Three-player combinations make this idea even harder to defend. Spain’s strength was not just one triangle, but the speed with which several overlapped. A midfielder could be one corner of multiple shapes, connected to a defender, full-back, winger and another midfielder.

Passing-network research describes players as points and passes as links. Yet a final network diagram for a given football match misses something important: Spain were rebuilding those links with every movement and first touch they made.

The value of a pass is not only where the ball goes, but how it changes what could happen next. That is why ten sideways passes are not automatically better than one pass that removes two defenders.

The shape that conquered basketball

Triangles are not unique to football. Under head coach Phil Jackson (1987-1998), US basketball team the Chicago Bulls made frequent use of a tactic called Tex Winter’s triangle offence. This places three players on one side of the court to create passing, cutting and scoring routes.

The sport was different but the principle was the same: triangles preserve alternatives. Close one route and another remains; follow the ball and space opens elsewhere.

In their match against Spain, France’s problem was not simply a lack of running. Pressing works only when several defenders move together and close connected passing options.

After the match, Kylian Mbappé said there had been “a lack of communication” in France’s pressing. Spain kept moving before France could reorganise.

A triangle on a tactics board does nothing by itself. Its power depends on players scanning before receiving, opening their bodies, and using the first touch to preserve the option of more than one next action.

The deeper contest was between Spain’s speed of reconfiguration and France’s speed of recognition. Spain kept changing the problem faster than France could solve it. The triangles were visible – but their real advantage was control over what could happen next.

Will we see a similar approach deployed in the final? Argentina will no doubt be closely studying their opponent’s tactics from other matches. But geometry will still play a prominent role – and may even prove a deciding factor.

Conor Boland does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Digital icons are everywhere – and usually, we interpret these visual symbols in the blink of an eye. Who today doesn’t know that a wastebasket means “delete” and a magnifying glass means “search”?

Yet icons’ meanings can be deeply shaped by culture, experience and technological history. So what seems obvious to one person may be confusing to another.

They were originally designed to make computers look friendlier and less intimidating to the few people who had access to them. For example, the icons below first appeared in 1975 in the Pygmalion visual programming system.

Years later, they attained much wider visibility thanks to the Xerox Star 8010, a pioneering office computer that introduced many graphic interface concepts that are still used today.

Research my colleagues and I conducted examines how people perceive, understand and evaluate digital icons – including which visual characteristics make icons easier to learn and use. Pygmalion’s pioneering icons worked because they relied on existing knowledge of the office world.

This meant they reduced cognitive effort by relying on recognition rather than recall. The icons looked like the physical objects they were representing – and critically, objects the computer’s users were already familiar with.

Five secrets of a good icon

The designer Susan Kare, creator of many of the original Macintosh icons, said a good icon should either be instantly recognisable or so easy to remember that a user only needs to learn it once.

To Kare, designing icons was about solving “the little puzzle of making an image fit a metaphor”. More than 40 years later, that challenge remains.

Research, including work by my colleagues and me, suggests the most successful icons tend to share key characteristics that guide us from seeing to understanding almost instantaneously. Here are five of them.

1. They depict things we already know

Early computer icons worked because they borrowed from the office world that people already knew: folders, bins, documents, calculators, floppy disks. Psychologists refer to those as “concrete” icons because people to use their knowledge of the everyday world to interpret them.

However, only a limited number of functions can be represented as identifiable objects, and getting a close fit between pictures and functions is not always easy. The more complex the meaning becomes, the harder it is to design concrete icons.

To test this, can you guess the meaning of these four icons? (Answers at the end of the article.)

2. They mean what we think they mean

Psychologists talk about “semantic distance” – how closely an image matches its intended meaning. An abstract symbol for “privacy settings” or “cloud syncing” has a much larger semantic distance than using a bin to mean “delete”.

As digital functions become more complex, designing icons that communicate their meaning quickly becomes increasingly difficult.

3. They feel familiar

Another important feature of successful icons is consistency of use over time – which leads to familiarity. The icons shown below are widely used even though the objects depicted are no longer so widely in use. This highlights the point that icons are partly informative signs and partly shared learned conventions, whose success is based on collective familiarity.

Take the floppy disk “save” icon (below left). Younger users recognise the meaning without ever having seen the physical object that the icon originally represented. The same is true of traditional telephone handsets and perhaps even envelopes (now widely used to denote “email”). While the objects have been superseded, the icons remain.

4. They look good

A well-designed object can have a positive effect on our behaviour – and the digital world is no different. Well-designed icons are more likely to attract downloads, help us perform tasks more efficiently, and learn better and faster. They even make digital environments feel more pleasant to use.

Think of the difference between an app icon that feels cluttered and amateurish and one that looks clean, balanced and professional. Even before we know what the app does, the icon’s design can influence expectations of how well the app will work.

5. They are tested with real users

Although icons and symbols are on the rise as a global visual language, it’s important not simply to assume that icons work globally – and to understand what makes a good digital icon across languages, cultures, ages and digital experiences.

This is why the International Standards Organisation (ISO 9186) demands comprehensibility testing – because symbols should be understood without explanatory text whenever possible.

A bridge between perception and meaning

As we spend more of our lives in digitally mediated environments, icons do two jobs simultaneously. They help us interact with technology more efficiently, and shape how we feel about the experience.

A digital icon is not simply a small picture. It is a bridge between perception and meaning. The best icons make interfaces feel less intimidating, more intuitive and more approachable – creating a global visual language that crosses barriers of language and culture.

In other words, good icons do more than help us find our way around a digital world. They help make that world feel understandable, welcoming and human.

Icon quiz answers: 1. cloud syncing; 2. privacy settings; 3. algorithmic recommendations; 4. generative AI.

Irene Reppa does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Europe is taking decisive steps to strengthen its control over the technology that underpins its societies. It is concerned about the digital infrastructure that sits behind everything from energy grids and water supplies to public transport, healthcare and the apps in people’s phones.

The European Commission has now published its package of measures for “tech sovereignty”, with the aim of building up its capabilities in digital technology.

The proposals rely heavily on private investment and may struggle to match global tech spending.

The plans include bolstering Europe’s semiconductor industry, expanding cloud computing, building up capabilities in AI, promoting open source software and setting out a plan for integrating digital technologies into energy systems. This last objective would support a sustainable expansion of data centres.

The backdrop is stark. Around 80% of Europe’s digital infrastructure and technology comes from outside the EU. An estimated €264 billion (£228 billion) a year is spent on cloud and software services that are mostly non-European.

Europe produces only about 10% of global semiconductors (the technology behind computer chips). In cloud computing, European providers hold roughly 15% of the market, while three US firms account for around 70%.

These imbalances were long seen as a natural outcome of globalisation. But in a world of geopolitical tension, they increasingly represent a strategic vulnerability. Export controls, sanctions and political pressure risk rippling through digital supply chains. The recent US ban on foreign access to advanced AI models sends a small yet clear signal.

What once looked efficient now looks risky. At the same time, the EU stresses that this is not about isolationism.

Hidden dependencies

A central shift proposed by the European Commission’s digital sovereignty package is to make digital dependencies visible and measurable. Through tools such as sovereignty risk assessments for cloud and AI services, organisations are pushed to look beyond performance and price and at the level of “sovereignty” a service or product provides.

Questions of ownership, jurisdiction, supply chains, data location and the ability to switch providers here become central when procuring digital services. The Computer & Communications Industry Association, which represents tech companies such as Amazon, Meta and Google, said the plans would undermine the EU’s goals of bolstering its digital infrastructure. On the other hand, European counterparts say the language in the tech sovereignty plans is soft.

This changes the role of buyers – especially in the public sector. Governments would no longer be just customers. Instead, they would start to shape markets. Through procurement and infrastructure investment, they can influence which technologies succeed and how digital ecosystems evolve.

However, open source software should enable re-use, transparency and compatibility between different IT systems. It provides an alternative to closed, vendor-controlled systems. It allows organisations and governments to collaborate on shared building blocks rather than repeatedly purchasing proprietary solutions from tech companies.

Public sector demand for open technologies has been shown to stimulate competition and economic growth. In this sense, the sovereign technology agenda is not just about building alternatives but about reshaping the rules of the digital economy.

The EU is trying to give organisations more freedom to switch between different technology providers and negotiate with them. This shifts the control back towards users – helping ensure continuity even in times of crisis.

---

Read more: Europe wants to end its dangerous reliance on US internet technology

---

Ambition faltering?

Turning this vision into reality will not be straightforward. Much of the content of the tech sovereignty plan relies on encouragement rather than binding measures, especially around open source software and procurement. This creates a risk that “sovereignty” becomes more of a political label than an operational shift, with uneven implementation across Europe.

Nor is this the first attempt at promoting a similar vision of change. Earlier initiatives – in the form of the Tallinn, Berlin and Strasbourg declarations, to more recent EU legislation – have outlined similar ambitions but these have struggled to materialise. Without clearer requirements, there is a risk of repeating this pattern.

Another challenge is fragmentation. If member states interpret sovereignty differently, the result could be a patchwork of approaches that weakens impact. Coordinated action is needed to grow both supply and demand across Europe. Public procurement will be central, but not sufficient. Organisations also need capabilities – the knowledge and structures to assess dependencies and act strategically.

Current frameworks tend to focus on evaluating technology providers, rather than whether organisations themselves can make and sustain sovereign choices.

This is where mechanisms like Open Source Program Offices (OSPOs) matter as promoted by the package. These support the governance and development of open technologies. Strengthening such capabilities may prove as important as policy itself.

The Commission’s funding levels for its tech sovereignty package raise further questions. Through an envisaged public-private (mostly the latter) funding model, around €2 billion (£1.7 billion) is calculated for open source over seven years, €120 billion (£103 billion) to semiconductors, €200 billion (£172 billion) to data centres – significant, but modest compared to the money invested in global technology platforms and the digital economy. Practical constraints remain: switching systems is costly, alternatives may lag behind, and shortages in skills and cybersecurity capacity add pressure.

National policies reflect these tensions. Sweden’s recent cloud computing strategy, for example, recognises the need to reduce dependencies but remains cautious in how far it goes – illustrating how national differences could limit Europe’s collective impact.

Whether the EU’s package of measures can succeed will depend less on ambition and more on implementation, and whether the supply and demand for sovereign tech is able to nudge. Negotiations to follow will show how things fall out.

Ultimately, this is not just a European issue, but one for any government or organisation. It is about who controls the digital systems modern societies depend on – and how those systems remain resilient, secure and aligned with public values in an uncertain world.

Johan Linåker does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

When disappointment strikes, is your instinct to try to shake it off, forget about it and move on? My research and experience of many workplaces suggests this might be exactly the wrong response.

My interest in the science of disappointment began more than 15 years ago as a workplace consultant. I was struck by how often clients described episodes that left them feeling disappointed as deeply personal and unsettling experiences – and by how little research there was to help me respond meaningfully. That prompted me to do a PhD on the subject.

Disappointment often reflects a gap between expectation and reality. It can involve grieving a future we had already begun to live in our minds.

My subsequent research with colleagues revealed a telling pattern. In the workplace, disappointment is frequently generated at a systemic level by unrealistic targets – yet lands on individuals as a sense of personal failure.

In many walks of life, it is commonly dismissed as an unwanted and unhelpful emotion. But our research tells a different story. Disappointment can be an important fuel for creativity. It surfaces what we truly desire, clarifies what matters to us, and points us toward what we are not yet willing to accept.

Whether in our professional or personal lives, disappointment is a signal worth learning to read. Here are some ideas for when you next come up against it.

1. Don’t get ahead of yourself

When we are waiting on a significant decision – a job offer, test result or relationship turning point – our emotional response is prepared long before the answer arrives. The same outcome can feel entirely different depending on what we anticipated would happen. The wider the gap between expectation and reality, the greater the disappointment.

In the workplace, severe disappointment in not getting a job or missing out on a promotion can stem from the loss of a working future we had already begun to imagine. If that future does not materialise, we grieve it – even if it never fully existed.

2. Beware the success trap

Success can quietly raise the bar for future failure. One of our respondents illustrated this dynamic neatly. Exceed your work target by 10% one year, they observed, and your manager is unlikely to reward you with a lighter load the next. Rather, the target is raised again, making falling short more likely – and the disappointment more acute because of your past success.

The same pattern can play out in social situations. Think of a friend who often picks up the bill. Over time, a generous gesture becomes expected behaviour. Then, on the one occasion they don’t pay, this becomes a moment of disappointment that people notice and remember. That disappointment is not proportionate to what actually happened, but to the gap with what was expected.

3. Try not to blame yourself (or anyone else)

People rarely experience disappointment in a neutral way. Rather, they tend to interpret it through one of two familiar patterns.

The first is internal: “I am the problem.” This assumes they did not try hard enough or were simply not good enough. Disappointment is treated as a sign they are a flawed or bad person.

The second interpretation is external. The fault is with others who did not recognise the person’s value and did not live up to expectations. The instinct is to blame and get angry with them.

Our research on disappointment in organisations shows both responses miss the point. Blaming ourselves or others can be a way of avoiding something harder to confront: that expectations are unrealistic or based on inaccurate assumptions.

4. The Ikea effect

Environments shape expectations. In workplaces, many people are encouraged to aim high and improve continuously. Organisations often promote ideals of progress, achievement and fulfilment.

These ideals can be motivating, but they can also create a perfect scenario that reality struggles to match. From this perspective, disappointment can be a structural feature of systems that rely on high expectations and idealised outcomes.

But there’s a personal aspect too. Research on what psychologists call “the Ikea effect” shows the more effort we invest in something, the more we value it – rather like a flatpack piece of furniture that we have built ourselves. At work, we routinely pour time, energy and identity into projects, roles and relationships. So, when things don’t go as hoped, we are losing something very personal.

And because failure at work is often witnessed by colleagues and managers, the stakes feel higher. The loss can become entangled with how others see us, and how we see ourselves.

Left unexamined, such feelings can calcify into something more damaging than the original disappointment: a diminished appetite for risk, a reluctance to invest fully in what comes next, and a growing suspicion that doing so is simply not worth it.

5. Be realistic, not idealistic

Moving from trying to eliminate disappointment to tolerating it can make it less destabilising and more informative. As a manager, this might mean developing the habit of noting, at the outset of a project, what a realistic rather than an ideal result would look like.

Similar patterns can appear in relationships too, where expecting things to feel perfect all the time can make an otherwise good relationship seem lacking.

Research consistently shows that naming difficult emotions reduces their intensity, and that workplaces where disappointment can be discussed honestly tend to be psychologically safer, more creative and better at learning from setbacks than those where such feelings are expected to be quietly moved past.

6. Accept disappointment, don’t dismiss it

Disappointment is uncomfortable because it confronts us with limits: to what we can control, to what organisations can deliver, or to what relationships can provide. An understandable instinct is to try to move past this quickly.

But a more constructive approach is to reflect on where our expectations come from, how they are formed, and whether they can be moderated in ways that benefit us. If disappointment is a signal that our expectations and reality are out of alignment, then understanding this may be one of the most important forms of resilience we can develop.

Annette Clancy does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.