These products all use play, but they return different things: an archetype, task scores, an entertainment profile, or a revisable working model.
The direct answer
TELLS, Human Benchmark, Refind Self, and Second Vision are not interchangeable personality tests. TELLS combines one set of mini-games into an archetype; Human Benchmark reports separate cognitive-task scores; Refind Self turns adventure choices into an entertainment personality result; Second Vision builds a revisable Persona only from registered evidence methods.
This comparison describes the contracts each publisher currently documents. It does not rank scientific accuracy. Research on game-related assessment finds substantial variation across tasks and validation designs, so sharing a game format does not make two products equivalent; see this meta-analysis of adult game-related assessments.
Factual comparison · reviewed 2026-08-22
Which play-based self-discovery format matches the result you actually want?
For: People choosing between a compact game archetype, separate cognitive-task scores, a narrative entertainment result, and a revisable multi-source working model.
Last reviewed .
| Criterion | TELLS | Human Benchmark | Refind Self | Second Vision |
|---|---|---|---|---|
| Input | Documented fact Its publisher presents 16 short behavioral mini-games and no questionnaire. Second Vision interpretationA compact game-only battery with one designed scoring model. | Documented fact Separate tasks cover reaction time, memory, sequence, typing, aiming, and related performance. Second Vision interpretationEach task is easier to read as a named performance score than as personality. | Documented fact An exploration adventure analyzes in-game routes, conversations, and choices. Second Vision interpretationThe longest-form and most narrative input in this set. | Documented fact The starting path combines one Bisect run, an immediate feeling response, and five reviewed self-view choices. Second Vision interpretationBehavior and self-description begin as separate clues rather than one hidden score. |
| Time and access | Documented fact Its publisher states about 12 minutes, no signup, and a result stored in a link. Second Vision interpretationThe lowest-friction option in this set for a single shareable session. | Documented fact Individual browser tasks are free to open; an account can retain dashboard history. Second Vision interpretationUseful when you want to repeat one task without completing a personality journey. | Documented fact The publisher says roughly one hour per playthrough and a conclusion after three; it is sold through digital stores. Second Vision interpretationA paid authored game experience rather than a short browser assessment. | Documented fact The starting preview takes about two minutes and is free before signup; an account saves the working model. Second Vision interpretationA short first read with a deliberate continuation boundary. |
| Immediate output | Documented fact One of 18 archetypes across nine named axes, including a rarity read. Second Vision interpretationA vivid categorical payoff rather than a source-separated evidence record. | Documented fact A score for each named task, with dashboard history and comparisons where available. Second Vision interpretationA cognitive-task benchmark, not an archetype or Persona. | Documented fact Up to five descriptions selected from 23 personality types, with a shareable result ID. Second Vision interpretationAn entertainment personality profile embedded in the game world. | Documented fact A bounded starting thesis with source, support, limitation, and an open question. Second Vision interpretationThe payoff is a challengeable working model, not a final type. |
| What repeated use changes | Documented fact The public product description presents one combined result from the mini-game set. Second Vision interpretationChoose it for a finished session read, not an indefinitely revisable profile. | Documented fact New attempts add score history for the same tasks. Second Vision interpretationRepeated use tracks performance change without synthesizing a wider self-model. | Documented fact Three playthroughs contribute to the designed conclusion. Second Vision interpretationRepeated play deepens that authored result rather than maintaining an open-ended evidence model. | Documented fact Later registered game, self-report, and optional outside-view evidence can support, qualify, or challenge a read. Second Vision interpretationChoose it when revision history matters more than a one-session label. |
| Evidence boundary | Documented fact The publisher links its games to named research paradigms. Second Vision interpretationParadigm ancestry does not by itself establish the validity of the shortened implementation or archetype mapping. | Documented fact The public site describes task performance rather than a validated personality assessment. Second Vision interpretationDo not promote a fast score into intelligence or character without separate evidence. | Documented fact PLAYISM explicitly calls the personality test entertainment rather than rigorous scientific assessment. Second Vision interpretationTreat the profile as part of the game, not a clinical or psychometric conclusion. | Documented fact Only registered methods can support Persona; other public games remain standalone results. Second Vision interpretationThe evidence boundary is narrower than the total playable catalogue. |
| Better fit when | Better fit when you want a quick, no-signup, game-only archetype experience. | Better fit when you want repeatable scores on named cognitive tasks. | Better fit when you want an authored adventure whose personality layer is entertainment. | Better fit when you want behavior and self-view kept separate inside a model designed to change. |
Limitation: The matrix compares documented product formats. It does not rank scientific accuracy, and it does not independently validate any product result.
Primary product sources: TELLS official product page · Human Benchmark official product page · Refind Self official product page · PLAYISM release and entertainment disclosure
TELLS: one multi-game archetype read
TELLS says its input is 16 quick mini-games adapted from tasks such as BART, the Iowa Gambling Task, go/no-go, digit span, and n-back. It maps choices onto nine named axes and returns one of 18 archetypes with a rarity read.
Its official page describes an approximately 12-minute, no-question, no-signup flow whose result is encoded in a shareable link rather than stored in a user profile. That is a session result; the page does not describe a longitudinal model that merges later evidence.
Human Benchmark: separate task scores and history
Human Benchmark offers separate reaction-time, sequence-memory, aim, number-memory, verbal-memory, visual-memory, typing, and chimp-test tasks. The immediate output is a task score, not a personality archetype.
Its account dashboard can show per-test scores, percentiles, and an activity feed across attempts. New attempts can therefore add performance history, but the documented output remains a set of task benchmarks rather than an evolving interpretation of personality.
Refind Self: an adventure-shaped entertainment profile
PLAYISM describes Refind Self as an exploration adventure that analyzes actions such as where you go, whom you speak with, and which activities you choose. It selects up to five descriptions from 23 personality types and issues a result ID that can be compared with another player’s result.
The publisher says the full test concludes after three roughly hour-long playthroughs and explicitly calls it entertainment rather than rigorous scientific assessment. Repeated play contributes to that designed conclusion; the official material does not describe an indefinitely evolving cross-context profile. It is sold through listed digital storefronts, whose availability and prices can change.
Second Vision: a working model with evidence boundaries
Second Vision currently begins with a Bisect run, an immediate feeling response, and five reviewed self-view choices. Its broader input model separates registered game behavior, registered self-report, and optional outside perspective.
The output is a starting thesis inside Persona, not a final type. The working model can change when later registered evidence adds support, creates a contradiction, or leaves a question open. Public games are free to play, an account saves the starting model, and tokens are used only when a user authorizes deeper Persona refinement.
How to choose without over-reading the result
Choose Human Benchmark when you want repeatable scores on named cognitive tasks. Choose Refind Self when you want an authored adventure whose personality layer is explicitly entertainment. Choose TELLS when you want one compact, game-only archetype experience. Choose Second Vision when you want to revisit a working interpretation as different registered sources are added.
Those are product-format differences, not proof that any result describes the whole person. Before trusting a psychological conclusion, inspect the exact task, scoring rule, reference sample, construct evidence, limitations, and whether the result is a one-session output or a model designed to change.