Users apply the model of one interaction style to a product in a different modality

What it is

Cross-product transfer isn't limited to a single interaction modality — touchscreen apps borrowing gestures and layouts from one another. It also happens across entirely different modalities, which is cross-modal transfer. Users carry inference rules formed in graphical interfaces straight over to voice assistants, wearables, or spatial-interaction products, even when the new modality has no physical way of supporting the original rule. A typical case: expecting a voice assistant to "remember what I just said and understand the context" — an expectation lifted directly from the GUI experience of a system retaining an operation history.

Why it happens

What a user's mental model encodes is usually an abstract interaction principle — "the system should preserve state," "an action should be undoable," "there should be somewhere to look back at what happened" — rather than a specific button, gesture, or screen layout tied to one modality. These abstract principles aren't inherently bound to the graphical form in which they were learned, so when a user enters a new modality where they have no independent experience yet, they default to filling the expectation gap with existing abstract principles — even when the new modality's physical constraints make no provision for how those principles could be realized. Voice has no screen surface to render a "history" view, but the underlying principle — "I should be able to check what was said before" — still gets carried over regardless.

Studying it

• Paradigm: have users with extensive GUI experience but no corresponding experience in a new modality complete a task on a new-modality product for the first time, and record the expectation-laden statements or help-seeking behavior they produce along the way. Map each expectation back to the GUI interaction principle it corresponds to, to see which principles get carried over most often.
• Methodological caveat: users of a new modality are accumulating experience quickly (voice-assistant users grow year over year), so the strength of cross-modal transfer weakens as the overall user base builds up its own experience in the new modality. Measurements from the same paradigm taken at different points in time shouldn't be compared directly across years.

Where it stops holding

• What transfers is the abstract interaction principle, not a modality-specific motor action — a GUI-specific gesture like "double-click" doesn't get carried into voice interaction, but principle-level expectations such as "the system should support undo" do, independent of how undo would actually be executed.
• For users who have already built up substantial independent experience in the new modality, the original modality's inference rules gradually get replaced by rules the new modality itself has cultivated. Cross-modal carryover is concentrated in the early stage of a user's exposure to the new modality.

Applying it

• Before designing a product in a new modality, list the abstract interaction principles the target users have already internalized from their primary source modality (usually GUI), and check item by item whether the new modality can realize the corresponding function. For principles that can't be realized, state the limitation explicitly early in the interaction, rather than letting users discover the gap only after being let down.
• How to check: after new users complete a typical task, ask "what did you expect it to do here," and classify the answers against specific GUI interaction principles — this surfaces which functions in the new-modality product are most likely to disappoint expectations and most in need of being stated upfront.

Related

• Same group: A7.13.1 Users transfer platform-level and brand-level conventions with different strength · A7.13.2 The first tool of a kind a user learns carries the greatest weight in later transfer · A7.13.4 Deliberately breaking a common convention needs a matching payoff, or the net loss exceeds the differentiation gain · A7.13.5 Similar appearance paired with different behavior produces the strongest negative transfer
• Nearby: A7.03 Metaphor lowers learning cost by borrowing existing knowledge
• Search terms: cross-modal transfer · modality transfer · voice interface expectations · mental model transfer