Diffusion of innovations in Axelrods model

Axelrod’s model for the dissemination of culture contains two key factors required to model the process of diffusion of innovations, namely, social influence (i.e., individuals become more similar when they interact) and homophily (i.e., individuals interact preferentially with similar others). The strength of these social influences are controlled by two parameters: $F$, the number of features that characterizes the cultures and $q$, the common number of states each feature can assume. Here we assume that the innovation is a new state of a cultural feature of a single individual – the innovator – and study how the innovation spreads through the networks among the individuals. For infinite regular lattices in one (1D) and two dimensions (2D), we find that initially the successful innovation spreads linearly with the time $t$, but in the long-time limit it spreads diffusively ($\sim t^{1/2}$) in 1D and sub-diffusively ($\sim t/\ln t$) in 2D. For finite lattices, the growth curves for the number of adopters are typically concave functions of $t$. For random graphs with a finite number of nodes $N$, we argue that the classical S-shaped growth curves result from a trade-off between the average connectivity $K$ of the graph and the per feature diversity $q$. A large $q$ is needed to reduce the pace of the initial spreading of the innovation and thus delimit the early-adopters stage, whereas a large $K$ is necessary to ensure the onset of the take-off stage at which the number of adopters grows superlinearly with $t$. In an infinite random graph we find that the number of adopters of a successful innovation scales with $t^\gamma$ with $\gamma =1$ for $K> 2$ and $1/2 < \gamma < 1$ for $K=2$. We suggest that the exponent $\gamma$ may be a useful index to characterize the process of diffusion of successful innovations in diverse scenarios.

💡 Research Summary

This paper investigates the diffusion of a single innovation within Axelrod’s cultural dissemination model, a framework originally designed to capture the interplay of homophily (the tendency to interact more with similar others) and social influence (the tendency to become more similar after interaction). The authors introduce an “innovator” who possesses a novel cultural state (denoted q + 1) on one of its F features, while all other agents start with randomly assigned states from the usual q possibilities. The innovator’s novel state is fixed; it cannot be altered by subsequent interactions, but any neighbor that adopts it may later lose it through normal Axelrod dynamics. The main observable is the mean number of adopters N ξ(t) − 1 (excluding the innovator) as a function of time t.

Model and Parameters
Each agent is characterized by F cultural features, each taking one of q possible states. The interaction rule is the classic Axelrod one: a pair of neighboring agents is selected at random, they interact with probability equal to their cultural similarity (the fraction of shared features), and if they interact a randomly chosen differing feature of the target agent is set equal to the neighbor’s state. The network topologies examined are (i) one‑dimensional rings, (ii) two‑dimensional tori, and (iii) Erdős‑Rényi random graphs with average degree K = 2 (critical) and K = 40 (highly connected). The innovation is introduced at time t = 0 at a single node (the origin).

Short‑time Dynamics
Across all topologies the number of adopters grows linearly at the very beginning: N ξ(t) ≈ 1 + v t, where the rate v depends on F and q but not on system size. The authors derive an exact expression for v by considering the probability that the innovator’s neighbor adopts the new state during the first update step. Larger q reduces v because agents share fewer features on average, making the initial interaction less likely.

Long‑time Behavior on Infinite Lattices

• 1D Ring: After the ballistic regime, the spread becomes diffusive. In the thermodynamic limit (L → ∞) the number of adopters scales as t^{1/2}. This mirrors the behavior of a voter model with a zealot, where the domain of the new state performs a random walk with diffusion constant D that can be estimated analytically.
• 2D Torus: The asymptotic growth follows a sub‑diffusive law N ξ ∼ t / ln t. The logarithmic correction arises from the slower coarsening dynamics in two dimensions, where domain boundaries are more constrained.

In both cases, if the underlying Axelrod system without an innovator would end in a multicultural frozen state (large q, small F), the innovation never reaches the whole population; it remains confined to a finite cluster even on an infinite lattice.

Random Graphs and the γ Exponent
On finite random graphs the classic S‑shaped adoption curve (slow start, rapid take‑off, saturation) can be reproduced, but only when the average degree K is sufficiently large. The authors argue that a trade‑off between K and q determines the shape: high q slows the early linear phase, while high K accelerates the take‑off. In the limit of infinite random graphs they find a power‑law growth N ξ ∼ t^{γ} with:

• γ = 1 for K > 2, independent of F and q (provided the system is in the consensus regime).
• ½ < γ < 1 for the critical case K = 2, where γ varies with q and F.

Thus the exponent γ encapsulates the combined effect of network connectivity and cultural diversity on diffusion speed. The authors propose γ as a universal index to compare diffusion processes across disparate social settings.

Implications and Future Directions
The study bridges cultural‑diffusion models and classic innovation‑diffusion theory (Rogers, 1962), showing that homophily and social influence alone can generate the familiar S‑curve under appropriate network conditions. The γ exponent offers a concise metric for policymakers: γ≈1 signals an environment where an innovation can spread almost unhindered (high connectivity, low cultural fragmentation), whereas γ < 1 warns of structural or cultural barriers. Extensions could include multi‑feature innovations, mutable innovators, external noise (cultural drift), or empirical validation with real‑world diffusion data.

Conclusion
By embedding a fixed‑state innovator into Axelrod’s model, the authors systematically characterize how lattice dimensionality, network degree, and cultural parameters shape the temporal evolution of adoption. Their analytical and numerical results reveal a universal early linear regime, a topology‑dependent long‑time scaling (diffusive, sub‑diffusive, or power‑law), and introduce the γ exponent as a novel, topology‑sensitive indicator of diffusion efficiency. This work enriches the theoretical toolbox for studying cultural and technological spread in complex social systems.