🔬 physics

The Attention-Directing Ability of Teams

This study introduces and validates "Attention-Directing Ability" (ADA) as a latent team capability derived from interaction patterns, demonstrating that teams with higher ADA are more effective at mobilizing collective effort and achieving superior performance in distributed innovation contexts.

Original authors: Olga Kokshagina, Marc Santolini, Christoph Riedl

Published 2026-07-30

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Olga Kokshagina, Marc Santolini, Christoph Riedl

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are watching a group of friends try to build a giant sandcastle on a beach. Some groups seem to have magic: one person shouts "I need a bucket!" and suddenly three people are running to get one, someone else is digging a moat, and the whole thing comes together in perfect harmony. Other groups, even with the same number of people and the same buckets, seem to fall apart. One person yells, and nobody hears them; another person digs a hole, but no one knows why. For decades, scientists studying how teams work have asked: What makes the difference? Is it who is in the group? Do they have the same plan in their heads?

This paper dives into a specific corner of science called organizational behavior, which studies how people work together. It builds on two big ideas. First, the idea of attention: just like your brain can only focus on one thing at a time, a team can only focus on a few tasks at once. The "Attention-Based View" suggests that how a team decides what to focus on determines if they succeed. Second, the idea of emergence: this is when a big, complex result (like a working team) pops out from tiny, simple actions (like sending a text message) that happen over and over. The paper asks a simple but tricky question: Is a team's success just about how much they talk, or is there a hidden "superpower" in how they talk that makes everyone else jump into action?

The Team's Hidden Superpower: The "Attention-Direction" Skill

The researchers, Olga Kokshagina, Marc Santolini, and Christoph Riedl, went hunting for this hidden superpower in a massive, real-world experiment. They looked at the OpenCovid19 Initiative, a global, online community where thousands of volunteers (2,233 of them!) came together during the pandemic to solve problems. These volunteers formed 79 different teams, chatting in 165 different "rooms" on a messaging app called Slack. They sent over 30,000 messages, trying to build tools, share ideas, and win small grants to fund their projects.

The team wanted to know: Why do some of these online groups click, while others just buzz with noise? They introduced a new concept called Attention-Directing Ability (ADA). Think of ADA not as a person's brainpower, but as a team's "musical conductor" skill. In a band, the conductor doesn't play every instrument; they wave a baton to tell the violinist when to play loud and the drummer when to stop. A team with high ADA is like a band with a great conductor: when one person sends a signal (like a question or a mention), it doesn't just disappear into the void. Instead, it triggers a chain reaction. A "ping" leads to a "reply," which leads to a "new idea," which leads to "more work."

How They Measured the Magic

To find this ability, the researchers didn't just count how many messages were sent. That would be like counting how many notes a band played without listening to the music. Instead, they used a super-precise method to watch the flow of time. They looked at every single message, reaction, and "ping" in 1-hour chunks. They asked: "When Person A sends a specific type of signal, does Person B (or C, or D) reliably respond in the next hour?"

They built a giant map of cause-and-effect. They found that in some teams, signals acted like a pebble dropped in a pond, creating ripples that grew into waves of activity. In other teams, signals were like throwing a pebble into a swamp; it sank immediately with no ripples. By analyzing these patterns across 2,233 people, they discovered that this "ripple-making" ability is a real, measurable trait. They called it Attention-Directing Ability (ADA). It's a latent (hidden) skill that a team possesses, just like a team might have a hidden talent for teamwork.

The Big Findings: It's About the Spark, Not the Volume

The study found three main things, and they are pretty cool:

The Skill is Real and Stable: Teams with high ADA didn't just get lucky for one hour. They kept this skill over time. As the teams worked together longer, their ability to direct attention actually got more stable. It's like a sports team that learns to anticipate each other's moves; the more they play together, the better they get at the "conductor" skill.
It Predicts Success: Teams with higher ADA were much more likely to finish a project and submit it for a grant. Even more importantly, when they did submit, their projects got higher scores from experts. It wasn't just that they talked more; it was that their talking worked.
The Secret Sauce is Effort: This is the most interesting part. The researchers found that ADA doesn't magically make a project better. Instead, it acts like a spark plug. High ADA makes the team work harder and longer. Because the signals kept triggering responses, the team stayed engaged, sent more messages, and put in more collective effort. That extra effort is what led to the better scores.

What It's NOT (And What We Can't Say Yet)

It's important to know what this study didn't find. The researchers explicitly ruled out the idea that it's just about how much a team talks. A team could send a million messages and still have low ADA if those messages didn't trigger any responses. It's not about the volume; it's about the connection.

Also, the paper doesn't claim that ADA is the only thing that matters. They didn't say, "If you have high ADA, you will definitely win." They found that ADA strongly predicts success and explains why some teams succeed where others struggle, but it's part of a bigger picture. They also didn't test this in a lab with actors; they used real data from real people working on real problems during a global crisis, which makes the findings very strong.

The Takeaway for the Rest of Us

So, what does this mean for you? If you are part of a group project, a gaming clan, or a sports team, the paper suggests that the secret to winning isn't just having the smartest person or the most tools. It's about creating a culture where your signals get picked up. When you ask a question, does someone answer? When you share a link, does someone look at it?

The authors suggest that in our modern world, where we often work with people we've never met in person, this "Attention-Directing Ability" is a superpower we can all try to build. It's about learning to send signals that don't just vanish, but actually pull the whole team into action. It turns a group of individuals into a single, coordinated machine that can tackle big, scary problems together.

Technical Summary: The Attention-Directing Ability of Teams

Problem Statement
While extensive research exists on team composition and shared cognition, a critical gap remains in understanding how modern, distributed teams coordinate action in the absence of formal structures. Existing literature often operationalizes collective attention as a static quantity (e.g., the number of people focusing on a topic) or as a cognitive state. However, this fails to capture the dynamic, emergent process by which teams mobilize effort and achieve performance. The authors posit that the differentiator between successful and struggling teams is not merely the volume of communication, but the efficiency with which micro-level interaction signals (e.g., mentions, replies, onboarding cues) trigger downstream engagement and coordinated responses. The central empirical challenge addressed is distinguishing whether temporal patterns in team interactions are merely situational fluctuations or evidence of a stable, latent team capability.

Methodology and Data
The study utilizes high-resolution digital trace data from the OpenCovid19 Initiative, a global, open-source collaboration launched in response to the COVID-19 pandemic.

Dataset: The analysis covers 2,233 participants collaborating asynchronously across 79 self-organized teams in 165 public Slack channels, generating over 30,000 messages (specifically 29,218 coded action instances across 13,912 unique events).
Operationalization of Attention-Directing Ability (ADA): The authors develop a novel three-step analytical pipeline to derive a latent measure of ADA:
1. Causal Responsiveness Estimation: Using distributed-lag negative-binomial mixed-effects models on hourly time-series data, the authors estimate the causal effect of every ordered pair of action types (e.g., does a "dyadic mention" at time t predict "replies" at time t+1?). This yields random slopes for each team and grant period, capturing how strongly specific signals mobilize responses relative to platform-wide activity.
2. Responsiveness Profiling: For each team-period, the positive random slopes are aggregated at the action level to create a "responsiveness profile," summarizing how effective each class of behavioral signal is at eliciting downstream engagement.
3. Latent Factor Extraction: Principal Component Analysis (PCA) and Exploratory Factor Analysis (EFA) are applied to these profiles. A single dominant latent factor is extracted, interpreted as Attention-Directing Ability (ADA). This factor represents the synergistic architecture where a team's ensemble of signals reliably mobilizes collective engagement.
Outcome Variables: Team performance is measured via a two-stage process: (1) Grant Submission (binary decision to submit a proposal) and (2) Evaluation Score (average reviewer rating of 0–5 for submitted proposals).
Analytical Strategy: To address non-random selection into the submission process, the authors employ a Heckman selection model (Tobit-2 specification). To test the mechanism, they conduct causal mediation analysis using communication effort (message volume) as the mediator.

Key Results

Existence of a Latent Capability: The analysis confirms a stable, dominant latent dimension underlying team attention dynamics. Teams differ significantly in their ADA scores, which exhibit temporal coherence (Spearman rank correlations between rounds range from $\rho = 0.46$ to $0.84$), increasing as teams accumulate shared experience.
Predictive Power of ADA:
- Submission: Higher ADA significantly predicts the likelihood of a team submitting a grant proposal ( $\beta = 1.42, p < .01$ ), even after correcting for selection bias.
- Performance: Conditional on submission, higher ADA significantly predicts higher expert-evaluated scores ( $\beta = 1.22, p < .01$ ).
Mediation Mechanism: Causal mediation analysis reveals that ADA operates almost entirely through collective effort.
- For submission, approximately 77% of ADA's total effect is mediated by message volume.
- For evaluation scores, approximately 89% of the effect is mediated by message volume.
- The direct effects of ADA on outcomes, controlling for effort, are statistically insignificant. This suggests ADA functions as an upstream coordination capability that sustains engagement, which in turn drives performance.
Robustness: Results hold across alternative temporal aggregations (1-day vs. 1-hour bins), different data inclusion windows, and alternative factor structures. Sensitivity analyses confirm that the mediated effects are robust to potential unmeasured confounding.

Key Contributions

Theoretical Extension of the Attention-Based View: The study shifts the locus of attention direction from organizational structures (top-down allocation) to emergent, peer-to-peer direction within teams. It reconceptualizes attention not as a scarce cognitive resource or a static state, but as a dynamic, measurable coordination capability embedded in interaction patterns.
Redefining Collective Attention: Moving beyond "amount of attention" or "co-attention," the paper introduces the concept of Attention-Directing Ability (ADA). This frames the critical difference between teams as the efficiency with which their signals recruit uptake and response, rather than the volume of attention they hold.
Process-Oriented Emergence: The study contributes to complexity theories of teams by demonstrating how macro-level capabilities emerge from micro-level interaction dynamics. It identifies ADA as a "compilation emergence" where the configuration of signals creates a responsiveness architecture greater than the sum of individual parts.
Methodological Innovation: The authors bridge the gap between micro-level digital trace data and macro-level team constructs. By preserving the directional and temporal structure of interactions to derive a latent causal measure, the study offers a template for measuring emergent team capabilities without relying on aggregate statistics or self-reports.

Significance and Claims
The paper claims that Attention-Directing Ability is a fundamental, stable team-level capability that explains why some distributed teams successfully mobilize collective effort while others struggle, even with similar resources. The significance lies in demonstrating that coordination is not just a structural or cognitive phenomenon but a relational one, enacted through the recursive process of signaling and responding.

The authors assert that ADA is not merely a reflection of communication volume but a distinct capability that determines a team's ability to transform individual actions into sustained collective trajectories. By identifying this capability, the study provides a predictive understanding of team dynamics in remote and hybrid work environments, suggesting that the ability to direct attention through interaction is a critical determinant of innovation and performance in complex, distributed settings. The findings imply that organizational success in such contexts depends on cultivating interaction patterns where signals reliably cascade into coordinated action.