Pianeta.Studio · Working Document · May 2026
Swarm Intelligence & Neural Prediction in Campaign Creative Development
Article outline — click each section to expand. The argument builds toward Pianeta's genuinely novel contribution at section 5.
01Definition and state of the artfoundationsespandi sezione
1.1
Swarm intelligence / collective behavioural prediction
Define the paradigm: swarm systems aggregate distributed signals from many agents to produce emergent collective predictions. Key players: MiroFish, Unanimous AI, Artificial Societies. Contrast with traditional individual expert judgement.
1.2
Computational neural prediction / in silico neuroscience
Define the paradigm: AI models trained on fMRI data predict whole-brain cortical responses to stimuli without needing real subjects. Anchor in TRIBE v2 (Meta/FAIR, March 2026) as the state-of-the-art tool. Distinguish from attention-prediction tools like Neurons Inc.
Sets the two methodological traditions up as distinct — this distinction matters because sections 2 and 3 argue they are complementary, not redundant.
02Who else is doing this?landscape
2.1
Neural prediction tools applied to creative
Cortex (TRIBE v2 demo wrapper, no case studies), Neurons Inc (attention/cognition, not whole-brain), Neurensics (real fMRI on storyboards, slow and expensive), Brainsight (predictive eye-tracking). All are post-production or single-test tools.
2.2
Swarm intelligence applied to advertising
Unanimous AI (Swarm platform, ad-reaction testing since 2017, clients include Disney and Credit Suisse), Artificial Societies (YC W25, 2.5M personas, Teneo case study), Limbik (synthetic audiences in Agent Cloud). All operate on the behavioural layer only.
2.3
The combined approach — sparse literature, no prior precedent
No published agency workflow combines swarm-based behavioural prediction and neural/computational prediction, applied iteratively during production. The combination of the two, used in parallel during the edit, has no prior precedent in published literature or documented agency practice.
The novel claim is not "we used TRIBE v2" or "we used a swarm" — both are now accessible. The novel claim is the workflow that runs them in parallel during the edit, before creative decisions are locked.
03The literature on combining behavioural and neural predictionacademic core
3.1
Where does the need come from?
Behavioural prediction asks what people will do with content (share, donate, ignore). Neural prediction asks what the brain processes during exposure (attention, emotional valence, social engagement). These are orthogonal — convergent evidence from both is stronger than either alone.
3.2
Behavioural + neural integration: published research
Anchor papers: Falk, Berkman & Lieberman (2012, Psychological Science) — neural focus groups predicted anti-smoking call volume better than self-report, with 2.8–32× lift. Knutson & Genevsky (2018) — neuroforecasting framework formalising NAcc/MPFC as population-level predictors. Falk et al. (2016, SCAN) — extending the framework to public health campaigns broadly.
3.3
Public interest and social cause campaigns
Genevsky, Västfjäll, Slovic & Knutson (2013, Journal of Neuroscience) — NAcc positive arousal, not guilt, drives donation lift from identifiable victim images. Zito et al. (2021, Frontiers in Psychology) — neuromarketing tools applied to UNICEF bequest campaigns at IULM Milan. NGO campaigns have heightened emotional processing and trust dependencies — neural prediction is especially relevant here.
This is the intellectual core. It shows Pianeta's approach is not invented — it extends an established academic lineage — but operationalises it in a way the literature has not yet documented.
04The state of the art in in silico neuroscience applied to advertisingtechnical context
4.1
TRIBE v2 and the gap between research and application
Released March 2026 by Meta/FAIR. Trained on 700+ subjects, predicts activity at ~70,000 cortical voxels — a 70× resolution jump over previous models. Meta frames it as a neuroscience research tool, not an ad-testing tool. No commercial agency has published a case study using it. The gap is wide open.
4.2
The NeuralBench framework (May 2026)
A standardised evaluation framework for neural prediction models released the same month as the ECLAG work. Positions Pianeta at the cutting edge of an actively moving field — not applying settled tools, but working at the frontier as the methodology is being formalised.
4.3
The commercial frontier
The neuromarketing market is valued at $4 billion in 2026 and growing rapidly. The structural shift from panel-based fMRI (expensive, slow, one-shot) to in-silico prediction (cheap, fast, iterative) is what makes Pianeta's approach viable now when it was not possible before.
Establishes that the tools are real, the market is real, and the timing is right. Sets up section 5 as the argument for why production-integrated use is the key methodological advance.
05The iterative production modelnovel contribution
5.1
The dominant post-testing paradigm and its problem
Standard neuromarketing runs after the edit is locked. For NGOs and public interest organisations, this means no guidance during development — neural prediction runs, the cut is complete, and then you get the result. By then, cost and time constraints make meaningful revision unlikely.
5.2
Pre-testing as a partial solution
Scholte, van der Leij & Lamme (2022) argued neuro testing should move to storyboard stage — earlier decisions have more leverage. But real fMRI at storyboard stage still costs €5,000–20,000 per session and takes days. It remains one-shot, not iterative.
5.3
The dual methodology approach
TRIBE v2 + MiroFish running in parallel during the edit. Each music track, each voice candidate, each video cut tested before the decision is made. Cost: GPU minutes and compute time, not scanner access. This is what the literature has described as desirable but has not documented as practiced — until the ECLAG campaign.
This is the payoff. All five sections lead here: the combined, iterative, production-integrated dual methodology is the thing no one has published. The ECLAG campaign is the first documented case study.
Main findings from literature
01
Neural prediction models can reliably forecast advertising attention and emotional resonance
02
Swarm-based collective prediction outperforms individual expert judgement substantially
03
The combination of behavioural and neural data adds predictive power beyond either alone
04
NGO and social cause campaigns have specific neurophysiological dynamics — heightened emotional processing and trust dependencies — that make neural prediction especially relevant
The core claim
No published workflow combines swarm-based behavioural prediction and computational neural prediction, applied iteratively during production. The ECLAG / Missing Children campaign is the first documented case. This is what makes Pianeta's approach genuinely stand out.
Continua a leggere
Altri articoli in arrivo.