Synergy IQ / The training network
Seven capacities.
One connected idea.
Seven training targets share one larger purpose: developing general intelligence. Start with an everyday challenge, practise the relevant cognitive skills and explore how learning across the network might support broader reasoning and learning. Each node also has an optional AI exercise coming soon.
Focused value now. Network value over time. Each node has a cognitive training target and an optional AI exercise coming soon. Practise a skill, apply it with tools and check what you can still do independently. Attention Control training is available now; the other nodes and all embedded AI exercises are in development.
How to read the network
Practise a skill.
Apply it. Check what stays.
The nodes combine established constructs with a small number of deliberately broader Synergy IQ training labels. Each node asks whether practice helps on a different task using the same capacity. The AI exercise coming soon asks another question: can you use assistance well while retaining understanding and independent skill?
The network asks a separate, harder question: can portable skills alter one another's learning trajectories and improve flexible problem solving? That developmental-leverage proposition is a testable Gf hypothesis, not a condition that must be met for one node to be worthwhile.
Attention Control
Find what matters. Resist competing signals.
“Everything pulls me away from what matters.”
Find the signal. Adapt your control. Recover when conditions change. Attention Control trains selective and interference control through a 12-block session: supported Capacity first, then Adaptive Control across Precision, Balanced and Speed objectives.
The majority-signal rule stays stable while later challenge sessions can change direction, introduce optic flow or add emotional distraction, followed by return to the original task. Optional Strategy Practice then asks: “What information actually matters here?” and can turn that into one IF/THEN plan for later review.
You are comparing three proposals while messages keep arriving. Attention control helps you keep the decision criteria active rather than being repeatedly captured by the newest signal.
Majority-signal selection under interference, supported Capacity and flexible control across different speed–accuracy objectives.
A held-out selective-attention or conflict task using different stimuli, rules and response mappings.
Keeping decision criteria active through notifications, noisy meetings and competing documents.
Improvement on the independent attention task would support near transfer—not better work performance, Gf or IQ.
How Attention Control training works and what the evidence supports → Available now — $20 →
AI suggests what matters in a mixed information stream. Check its choices against your goal and spot what it missed.
Choose the relevant signals yourself first; then compare the AI shortlist.
Check relevance and omissions on a fresh example without AI.
A proposed exercise to evaluate, not a released feature or a demonstrated training benefit. Cognitive training can be used on its own.
Why it matters for intelligent thinking
Attention control is increasingly treated as a coherent individual-differences ability. A 2024 meta-analysis and re-analysis of 90 samples and more than 23,000 participants found a broad attention-control factor related to working memory, fluid intelligence, long-term memory, reading comprehension and processing speed.[1]
paper
Unsworth, Miller & Strayer (2024)
Individual differences in attention control: A meta-analysis and re-analysis of latent variable studies.
Open DOI ↗Relational Memory
Keep the relationship active as information changes.
“I lose the whole picture when one part changes.”
Practise keeping relationships in mind and updating them as information changes.
A project depends on supplier A, which depends on approval B, which changes the timing of team C. You need the relationships—not just the individual facts—to stay active.
Maintaining and updating relationships among several items as the information changes.
A held-out relational-integration task using new materials and a different presentation format.
Tracking dependencies, changing assumptions and hand-offs across a complex project.
A gain on the new relational task would support construct transfer, not broader reasoning or project outcomes.
Build an editable map with AI, check how the parts connect and update it when a constraint changes.
Map the key relationships yourself before comparing an AI-assisted version.
Reconstruct the essential relationships without the map.
A proposed exercise to evaluate, not a released feature or a demonstrated training benefit. Cognitive training can be used on its own.
Why it matters for intelligent thinking
Relational integration is closely tied to fluid reasoning. Chuderski found that a task designed to measure the construction of complex relational structures was the strongest predictor of fluid reasoning and explained variance beyond four established working-memory tasks.[2]
paper
Chuderski (2014)
The relational integration task explains fluid reasoning above and beyond other working memory tasks.
Open DOI ↗Binding Memory
Keep features, roles and sources attached to the right item.
“I remember the claim, but forget who said it or what supports it.”
Practise keeping details, roles and sources attached to the right item.
You remember three deadlines, three clients and three constraints. Binding memory is what keeps the right deadline and constraint attached to the right client.
Keeping features, roles, locations and sources attached to the correct item under load.
A held-out feature-binding or source-memory task using new combinations and different materials.
Avoiding client, deadline, evidence-source and responsibility mix-ups when several cases are active.
A gain would support binding transfer; it would not establish a general working-memory or intelligence increase.
Use AI to organise claims and supporting passages, then check that each source actually supports its claim.
Match claims to their original passages and check people, roles and context.
Check source attribution on new material, including misleading matches.
A proposed exercise to evaluate, not a released feature or a demonstrated training benefit. Cognitive training can be used on its own.
Why it matters for intelligent thinking
Flexible relational bindings appear to be one component shared by working memory and fluid intelligence. Experimental work has also shown that working-memory load can reduce the strength of relational bindings and impair intelligence-test performance.[3] Related work argues that durable, flexible bindings help explain the common variance between working memory and Gf.[4]
paper
Schubert et al. (2023)
Working memory load affects intelligence test performance by reducing the strength of relational item bindings and impairing the filtering of irrelevant information.
Open DOI ↗Path Horizon
Track where a sequence came from and where it may go next.
“I can see the next step, but not where it might lead.”
Practise tracking sequences, dependencies and possible next states.
You can see that one project decision opens two possible routes, one of which creates a later bottleneck. You are mentally representing a short horizon of reachable states.
Representing ordered transitions, reachable future states and where a sequence has come from.
A held-out sequence-learning or graph-inference task using a new topology, symbols and transition structure.
Anticipating dependencies, bottlenecks and downstream consequences before committing to a route.
Because Path Horizon is an emerging construct, even the proposed near-transfer measure requires direct validation.
Compare AI-proposed paths with your own prediction. Identify constraints and decide what would make you change course.
Predict a sequence first; then compare alternatives and revision triggers.
Plan a new sequence independently and explain when you would revise it.
A proposed exercise to evaluate, not a released feature or a demonstrated training benefit. Cognitive training can be used on its own.
Why it matters for intelligent thinking
Path Horizon is a Synergy IQ training label, not an established IQ factor. Its scientific basis comes from work on cognitive maps, relational structure and inference. Tenderra and Theves found that fluid intelligence was positively related to the quality of map-like hippocampal representations, but not to ordinary non-relational item-memory signals.[5] A 2026 developmental study further linked maturing non-spatial cognitive-map codes to inference and real-world intelligence measures, particularly reasoning.[6]
paper
Tenderra & Theves (2025)
Human intelligence relates to neural measures of cognitive map formation.
Open DOI ↗Emerging extension: the exact −n/+n “horizon” construct used by Synergy IQ still requires direct validation and transfer testing.
Knowledge Access
Bring the relevant concept or rule into mind when it is needed.
“I understood it yesterday. Why can’t I bring it back now?”
Practise retrieving useful knowledge when the question or cue changes.
An unfamiliar problem suddenly becomes easier because you recognise that it has the same underlying structure as something you learned months ago.
Retrieving the relevant rule, concept or precedent under goal control and changed cues.
A held-out retrieval-and-application task requiring learned rules to be recovered from new cues after delay.
Bringing the right policy, model or prior case to mind when a pressured problem initially looks unfamiliar.
A gain could reflect better cue use or retrieval strategy; it would not show that factual knowledge or Gf increased.
Try recalling first, then use progressively stronger AI hints linked to approved learning material.
Attempt an answer before asking for a hint; check it against the learning material.
Return after a delay and retrieve the knowledge without hints.
A proposed exercise to evaluate, not a released feature or a demonstrated training benefit. Cognitive training can be used on its own.
Why it matters for intelligent thinking
Unsworth and colleagues decomposed the working-memory–fluid-intelligence relationship into capacity, attention control and secondary-memory retrieval. All three made shared and unique contributions to Gf, supporting the idea that controlled access to stored knowledge is part of intelligent performance.[7]
paper
Unsworth et al. (2014)
Working memory and fluid intelligence: Capacity, attention control, and secondary memory retrieval.
Open DOI ↗Generative Search
Generate alternatives before committing to the first answer.
“I keep coming back to the same idea.”
Practise generating and comparing alternatives before settling on an answer.
A system has failed. Rather than locking onto the first explanation, you generate several competing hypotheses and then ask what evidence would discriminate between them.
Producing multiple candidate explanations or solutions and switching between useful search categories.
A held-out divergent-search task in a new domain, scored for relevant fluency, flexibility and non-redundancy.
Generating competing incident hypotheses, strategic options or alternative interpretations before deciding.
This would test near transfer within generative search, not creativity in general or an increase in Gf.
Offer your own possibilities before AI adds alternatives, analogies and counterexamples. Judge which are useful.
Generate initial options independently, then look for genuinely different possibilities.
Try a new problem without AI; assess variety and usefulness, not just option count.
A proposed exercise to evaluate, not a released feature or a demonstrated training benefit. Cognitive training can be used on its own.
Why it matters for intelligent thinking
Generative Search is a broader Synergy IQ label rather than a standard intelligence factor. Its closest evidence base includes divergent thinking, retrieval fluency and executive search. Nusbaum and Silvia found that fluid intelligence predicted divergent-thinking performance and that part of this relationship was mediated by executive switching between idea categories.[8]
paper
Nusbaum & Silvia (2011)
Are intelligence and creativity really so different? Fluid intelligence, executive processes, and strategy use in divergent thinking.
Open DOI ↗Emerging extension: evidence linking divergent search to intelligent performance is not the same as evidence that divergent-thinking training raises Gf.
Reasoning
Combine relationships and evidence to infer what is not directly given.
“The answer sounds convincing. But does it actually follow?”
Practise combining relationships and evidence to test a conclusion.
You know that A causes B only under condition C. A new case contains A, but not C, so you resist the superficial analogy and conclude that B does not necessarily follow.
Combining relations, rules and constraints to derive and check conclusions from incomplete information.
A held-out relational, analogical or rule-inference task using unfamiliar content and problem forms.
Testing whether a conclusion follows, identifying missing conditions and resisting superficial analogies.
A gain on a different reasoning task would support construct transfer; one trained format may still remain narrow.
Ask AI to challenge assumptions and offer counterarguments. Check its evidence and decide what would change your mind.
State an initial judgement and confidence; test the strongest objection before deciding.
Solve a new problem independently and explain the evidence and uncertainty.
A proposed exercise to evaluate, not a released feature or a demonstrated training benefit. Cognitive training can be used on its own.
Why it matters for intelligent thinking
Reasoning is one of the clearest behavioural expressions of fluid intelligence. Neuropsychological work by Woolgar and colleagues showed that lesion volume within the domain-general multiple-demand system—but not the language system—predicted deficits in fluid intelligence.[9]
paper
Woolgar et al. (2018)
Fluid intelligence is supported by the multiple-demand system not the language system.
Open DOI ↗Why call it a network?
The nodes are meant to interact.
Attention can influence what enters and remains available to memory. Relational and binding processes help determine how information is structured. Knowledge access supplies prior concepts. Generative search expands the candidate space. Predictive mapping represents possible paths. Reasoning constrains what follows.
Synergy IQ’s stronger hypothesis is asymmetrical mutualism: improvement in some capacities may change how effectively another capacity can subsequently be learned. That developmental-leverage claim is not established by the studies above; it is a research hypothesis the programme is designed to make testable.
References
Bateman, J. E., & Birney, D. P. (2019). The link between working memory and fluid intelligence is dependent on flexible bindings, not systematic access or passive retention. Acta Psychologica, 199, 102893. https://doi.org/10.1016/j.actpsy.2019.102893
Chuderski, A. (2014). The relational integration task explains fluid reasoning above and beyond other working memory tasks. Memory & Cognition, 42(3), 448–463. https://doi.org/10.3758/s13421-013-0366-x
Nusbaum, E. C., & Silvia, P. J. (2011). Are intelligence and creativity really so different? Fluid intelligence, executive processes, and strategy use in divergent thinking. Intelligence, 39(1), 36–45. https://doi.org/10.1016/j.intell.2010.11.002
Qu, Y., Ou, J., Pang, L., Wu, S., Luo, Y., Behrens, T. E. J., & Liu, Y. (2026). Development of non-spatial grid-like neural codes tracks inference and intelligence. Cell, 189(10), 3091–3106.e15. https://doi.org/10.1016/j.cell.2026.02.028
Schubert, A.-L., Löffler, C., Sadus, K., Göttmann, J., Hein, J., Schröer, P., Teuber, A., & Hagemann, D. (2023). Working memory load affects intelligence test performance by reducing the strength of relational item bindings and impairing the filtering of irrelevant information. Cognition, 236, 105438. https://doi.org/10.1016/j.cognition.2023.105438
Tenderra, R. M., & Theves, S. (2025). Human intelligence relates to neural measures of cognitive map formation. Cell Reports, 44(8), 116033. https://doi.org/10.1016/j.celrep.2025.116033
Unsworth, N., Fukuda, K., Awh, E., & Vogel, E. K. (2014). Working memory and fluid intelligence: Capacity, attention control, and secondary memory retrieval. Cognitive Psychology, 71, 1–26. https://doi.org/10.1016/j.cogpsych.2014.01.003
Unsworth, N., Miller, A. L., & Strayer, D. L. (2024). Individual differences in attention control: A meta-analysis and re-analysis of latent variable studies. Psychonomic Bulletin & Review, 31(6), 2487–2533. https://doi.org/10.3758/s13423-024-02516-1
Woolgar, A., Duncan, J., Manes, F., & Fedorenko, E. (2018). Fluid intelligence is supported by the multiple-demand system not the language system. Nature Human Behaviour, 2(3), 200–204. https://doi.org/10.1038/s41562-017-0282-3
What changes across the network?
Look separately at learning, speed with accuracy, decision quality, resilience and broader cognitive development. Better AI-assisted output does not by itself show a change in your underlying abilities.