Figure 2: Three views of fMRI encoding (schematic).(a) Deterministic regression maps the stimulus 𝐱\\mathbf{x} to a single point estimate 𝐲^\\hat{\\mathbf{y}} (red), collapsing the target BOLD distribution onto its conditional mean. (b) Standard CFM draws source samples 𝐳0\\mathbf{z}\_{0} (orange) from a fixed Gaussian p0=𝒩(𝟎,𝐈)p\_{0}=\\mathcal{N}(\\mathbf{0},\\mathbf{I}) and transports each to pdata≈p(𝐲∣𝐱)p\_{\\text{data}}\\!\\approx\\!p(\\mathbf{y}\\mid\\mathbf{x}) along a long ODE path. (c) MultiFlow (ours) draws the *same* samples from a *stimulus-conditioned* source p0(𝐳0∣𝐱)=𝒩(𝝁ϕ(𝐱),σϕ(𝐱)2𝐈)p\_{0}(\\mathbf{z}\_{0}\\mid\\mathbf{x})=\\mathcal{N}\\!\\bigl(\\boldsymbol{\\mu}\_{\\phi}(\\mathbf{x}),\\sigma\_{\\phi}(\\mathbf{x})^{2}\\mathbf{I}\\bigr)whose 𝝁ϕ\\boldsymbol{\\mu}\_{\\phi} and σϕ\\sigma\_{\\phi} are *learned* to place the source closer to the target, shortening the expected transport path. In (b,c) the curves are independent samples from a *single* model; the panels differ only in the source distribution.## 2 Related Work 出處: MultiFlow: Vision-Driven Multimodal fMRI Encoding via Stimulus-Informed Source Flow Matching（arXiv）
Figure 1: Overview of the MultiFlow pipeline.Time-aligned *visual frames*, *audio samples*, and *transcripts* pass through ten frozen foundation-model backbones; the fused representation drives a generative model that synthesises whole-brain fMRI activity over the Schaefer-1000 parcellation. Rather than a single point estimate, MultiFlow models the stimulus-conditioned BOLD distribution via Stimulus-Informed Source Flow Matching (SISFM).Naturalistic paradigms in which participants watch movies during fMRI scanning have become a standard setting for studying how the brain integrates vision, language, and sound over time (Figure [1](https://arxiv.org/html/#S1.F1 "Figure 1 ‣ 1 Introduction ‣ MultiFlow: Vision-Driven Multimodal fMRI Encoding via Stimulus-Informed Source Flow Matching")). Encoding models that predict cortical responses from such stimuli give a principled way to probe this integration \[[Yamins and DiCarlo(2016)](https://arxiv.org/html/#bib.bibx33)\], and the Algonauts 2025 dataset \[[Gifford et al.(2025b)Gifford, Bersch, St-Laurent, Pinsard, Boyle, Bellec, Oliva, Roig, and Cichy](https://arxiv.org/html/#bib.bibx8), [Boyle et al.(2023)Boyle, Pinsard, Borghesani, Paugam, DuPre, and Bellec](https://arxiv.org/html/#bib.bibx2)\], with tens of hours of whole-brain recordings per participant, makes it feasible to train deep encoders on them. 出處: MultiFlow: Vision-Driven Multimodal fMRI Encoding via Stimulus-Informed Source Flow Matching（arXiv）
