hri-related-work

HRI Related Work

An HRI related-work section has a harder job than most: it must place the paper in **several

literatures at once** — robotics, HCI, psychology/cognitive science, and design — because HRI

reviewers are drawn from all of them, and each expects to see their field cited correctly. A section

that reads as an ICRA lit review to a psychologist (or vice versa) signals the authors do not

understand the interdisciplinary conversation they are joining. This skill covers coverage,

positioning, and double-blind hygiene.

Cover the crossing literatures

Map your paper to the lanes it actually touches, and cite each in that community's terms:

• HRI conference and journal core — the directly adjacent HRI work (prior HRI/THRI/JHRI studies

on your construct or interaction). This is where reviewers look first; missing the nearest HRI

neighbor is the classic fatal gap.

• Robotics — the technical substrate (perception, control, behavior generation, autonomy) your

robot relies on, if the contribution is technical/systems.

• HCI — interaction design, evaluation methods, and any general-HCI phenomenon your interaction

instantiates.

• Psychology / cognitive & social science — the theory behind your constructs (trust, mind

perception, social presence, warmth/competence, persuasion), cited to its origin, not only to its

HRI re-uses.

• Design — if yours is a design contribution, the design methods and precedents (probes,

research-through-design, participatory design) it builds on.

You do not need every lane — you need the lanes your contribution genuinely engages, covered well

enough that a reviewer from each field trusts you.

Delta-first positioning

HRI reviewers reward a crisp statement of what is new:

• For each closest prior work, state in one sentence what they did and **what you do

differently** — a different robot, a different measured outcome (behavior vs. attitude), a

different population, a causal manipulation where they had correlation, a design where they had a

study.

• Prefer a short, sharp delta over an exhaustive survey. Eight pages do not allow a textbook

review; they allow a targeted argument for why your contribution is needed.

• Avoid the "no one has studied X with a robot" claim unless it is true and load-bearing — novelty

of combination is fine, but say why the combination matters, not merely that it is unclaimed.

Credit theory correctly (the interdisciplinary trap)

The most common HRI positioning error is citing a psychological or design construct only through its

HRI re-uses:

• If you measure trust, anthropomorphism, social presence, or persuasion, cite the

originating theory and the validated instrument, not just the last three HRI papers that used the

word. A psychology reviewer notices immediately.

• If you borrow a method (thematic analysis, mixed methods, a specific experimental paradigm),

cite its methodological source so a methods reviewer trusts you are using it correctly.

• Do not rename an existing construct; align your terminology with the field that owns it.

Double-blind self-citation

HRI full-paper review is double-blind; positioning must not de-anonymize you:

• Cite your own prior work in the third person ("Prior work [X] found...", never "our previous

work [X]"), and keep those citations in the normal reference list.

• If a new submission builds directly on your unpublished or under-review work, describe the

relationship without identifying the authors; do not attach an identifying tech report.

• Do not remove necessary self-citations to avoid detection — a missing obvious reference is itself

a signal. Anonymize the phrasing, not the scholarship.

• Watch acknowledgments, grant numbers, and platform/lab names elsewhere in the paper that a

related-work reviewer might connect to an author (see hri-submission).

Placement and length

• A related-work section works when the paper needs to survey a landscape; inline positioning

(delta stated where each idea is introduced) often serves better under the 8-page budget. Choose

one primary strategy; do not both survey at length and re-argue inline.

• Keep related work proportional — it exists to justify the contribution, not to prove reading. If it

crowds out the method or results, cut it.

Anti-patterns HRI reviewers flag

• Single-field tunnel vision — a robotics-only or psychology-only reference list for a paper that

claims to be interdisciplinary.

• Construct laundering — using a theoretical term (e.g., "trust") while citing only its HRI

buzzword uses, not its theoretical source or validated measure.

• Survey-as-substitute — a long neutral catalogue with no delta, leaving reviewers to guess the

contribution.

• First-person self-citation — an anonymity break hiding in the related work.

Output format

[Literature lanes covered] HRI-core / robotics / HCI / psychology / design — which apply, which cited well
[Nearest HRI neighbor] identified and delta stated? yes/no
[Delta-first] each key prior work paired with an explicit difference? yes/no
[Theory credited] constructs/methods cited to origin + validated instrument? yes/no
[Double-blind] self-citations third-person; no identifying acknowledgments/labs? yes/no
[Fix queue] <ordered edits>