The Cognitive-Network Welfare Theorem

Broker-controlled information turns cognitive bias into a welfare cost

The paper joins Burt's structural-hole brokerage with Kahneman's cognitive architecture. When a broker spans a hole and frames information for biased recipients, total system loss includes a positive cognitive-network interaction term.

• Main equation: ΔW_total = ΔW_brokerage + ΔW_bias + ΔW_interaction
• ΔW_interaction > 0 when structural-hole depth h > 0 and cognitive distance d > 0
• The interaction rises with h and d because broker framing and biased processing are complements

This is a formal synthesis paper, not a simulation paper

The contribution is analytical: it states a welfare theorem, gives proof sketches, and calibrates the mechanism from existing empirical estimates rather than running a new Monte Carlo study.

• Tier 2 evidence: formal theoretical extension bridging two established empirical programs
• Method: Burt 1992 to 2015 plus Kahneman 1971 to 2021
• Structure: five propositions, three case studies, and an 8-outcome taxonomy mapping

Low constraint gives the broker timing, filtering, and framing power

The network primitive is Burt's constraint index c_B. Structural-hole depth is h = 1 - c_B, so deeper holes mean less redundancy and more broker control over cross-group information.

• Broker returns come from information benefits, control benefits, and referral benefits
• Bridged-party costs are filtering cost F(h), timing cost T(h), strategic framing cost S(h), and opportunity cost O(h)
• The prior Structural-Hole Welfare Theorem says broker returns rise linearly while bridged-party costs rise superlinearly

The recipient's benchmark is distorted before the message is evaluated

The cognitive primitive vector contains prospect-theoretic parameters, heuristics-and-biases parameters, System 1/System 2 parameters, and noise parameters. Cognitive distance d measures departure from the rational benchmark θ*.

• Loss aversion: λ ≈ 2.25, so closure costs feel larger than equivalent closure gains
• WYSIATI: System 1 builds a coherent story from available inputs without checking missing inputs
• Noise: cross-decision variance adds welfare loss beyond mean bias

WYSIATI lets a broker make an incomplete message feel complete

The broker selects, delays, and narrates raw states ω into messages m through f_B: Ω → M. The recipient receives a coherent story, while excluded facts remain outside the frame.

• The paper calls this WYSIATI-driven framing completion
• S(h) is amplified by the WYSIATI closure rate ρ(d)
• Gargiulo and Benassi: managers with one intermediary adopted that intermediary's frame in 78% of cases versus 34% with direct access, a 44-point gap

Loss aversion protects the broker from direct-tie closure

Creating a direct tie has upfront costs: time, social risk, conflict with the broker, and possible future information loss. The bridged party codes those costs as losses relative to the brokered status quo.

• The paper calls this loss-aversion bridge avoidance
• With λ ≈ 2.25, the perceived loss from closure can exceed the perceived gain even when objective welfare improves
• The monopoly barrier is cognitive: the broker controls the information that would make closure attractive

Availability lets the broker move attention away from slow system damage

The broker can make selected risks vivid, recent, and emotionally accessible while the gradual degradation of the information system stays abstract.

• The paper calls this availability-driven attention redirection
• The broker acts as a strategic availability engineer
• The recipient reacts to broker-selected crises while missing the welfare cost of dependence on the broker

The interaction term is positive because framing and susceptibility are complements

The theorem uses six axioms: brokerage control, strategic framing, cognitive susceptibility, interaction complementarity, system-welfare separability failure, and an admissible intervention domain.

• Reduced form: ΔW_total(h,d) = α0 + α1h + α2d + α3hd + ε
• α1 > 0 is pure brokerage cost, α2 > 0 is pure bias cost, and α3 > 0 is the cognitive-network interaction
• Lower bound: ΔW_CN ≥ h·d·[κρ·ρ(d) + κλ·(λ(d)-1) + κa·a(d)]

The three cases pair one broker position with one exploited bias channel

The case studies are not loose analogies. Each maps a brokered market or platform onto h, d, the broker's framing strategy, and a specific welfare leakage channel.

• Credit rating agencies: anchoring plus information monopoly between issuers and investors
• Social media platforms: System 1 exploitation plus algorithmic gatekeeping between producers and consumers
• Pharmacy benefit managers: loss aversion plus spread extraction between drug manufacturers and health plans

The default outcome is a (0,1,1) Hollow Win

In the Missing System taxonomy, the cognitive-network interaction makes all local parties feel better while the system loses. The paper codes this as system C = 0, bridged parties A = 1, broker B = 1.

• Broker gains private control payoff from the structural hole
• Bridged parties subjectively gain because broker-framed information feels useful and complete
• System welfare falls because filtering, timing, framing, opportunity cost, and cognitive distortion are not captured in the bilateral payoff

Policy has to reduce both h and d on the relevant support

The theorem's policy result is narrow: debiasing alone can be reframed by the broker, and closure alone can fail when recipients process new information through the same distorted frame.

• Independent verification adds a system-welfare signal outside the broker's channel
• Joint intervention combines direct ties, competing frames, structured decision protocols, and noise audits
• Paper metadata sets β_W reference floor = 5.89 and Π reference sign = negative