Choose two of four actual study designs. The comparison keeps the population, task, recorded response and prior exposure apart. Six pair-specific notes explain why the results cannot become one intelligence ranking.
READ THE CONNECTION
Full comparison as plain text
What the comparison contributes
The rows are an original editorial crosswalk, not pooled data or a meta-analysis. The pairing note is specific: the 2024 and 2026 papers share participants, whereas the other pairs differ in selection, task or exposure. None of the sources reviewed this tool. Their findings remain in the papers, not in a new composite measure.
What stays local
Only fixed study choices are used. There is no free text, upload, animal profile, browser storage or tool submission. Reload resets the pair. Optional copying includes the full comparison, source links and limits, and reports failure when access is unavailable.
1,002 dogs aged one to eight years in Finland; twelve breeds plus a mixed-breed category, selected through a commercial test battery. Numbers vary by task.
Task
Several tasks, including human gestures, a transparent barrier, short-term memory and an unsolvable problem.
Recorded response
Different outcomes for each task, such as correct choices, time to solve and time directed toward people. Some breed differences were found; not for memory or logical reasoning in this sample.
Exposure and time
A test-battery session, not the later UBC follow-up. Most individual training histories were unavailable.
What this does not establish
No universal breed IQ, individual rank or isolated genetic cause. A null finding does not prove identical ability.
111 final task-completers in five breed clades. Owners took part with researchers guiding them remotely.
Task
A hand-touch task with discrimination, reversal and extinction phases. Many dogs already knew hand touching.
Recorded response
Task-derived learning and behavior components. Differences depended on the phase and clade; this was not a global intelligence measurement.
Exposure and time
Initial testing reported in 2024. The 2026 publication follows dogs from this cohort, not an independent sample.
What this does not establish
Small clade samples, grouping, social context and uncorrected exploratory multiple comparisons limit interpretation. No home-test or breed-ranking claim.
105 dogs from the earlier 111 first-test completers in the UBC hand-touch cohort.
Task
The hand-touch task on two occasions about thirty days apart.
Recorded response
Average reversal performance improved; comparable overall improvement was not found for every component. The magnitude of change did not significantly differ among clades.
Exposure and time
Repeated exposure, with possible owner practice between tests not controlled. Sequential phases and only two occasions also limit interpretation.
What this does not establish
Not independent replication of the 2024 cohort or proof that practice caused a general intelligence gain. Non-significant differences are not exact equality.
The Finnish battery compares individual breed categories across several tasks. The UBC hand-touch work compares clades within one social learning sequence. A shared word such as learning does not put the responses on the same scale. You cannot add or rank their results as a single breed intelligence score.
A broad battery and a rare vocabulary sample
The Finnish dogs were selected through a commercial battery; the word learners were selected for an unusual existing skill. Their tasks and recorded responses differ. The seven full-completer result cannot tell you how many typical dogs would extend toy labels, and it cannot rank those dogs against the Finnish sample.
A task battery is not a repeat-exposure result
The Finnish study reports several task outcomes in its selected sample. The later UBC study follows change across two occasions in a different cohort and task. Higher performance in one outcome and change over time are different quantities. Neither can be used to supply a missing global IQ for the other.
Changing a response and extending a label
The hand-touch sequence and functional-label task ask different questions, in differently selected samples. Both involve people, but that does not make response updating and object-label extension interchangeable. The comparison cannot reveal which group is generally smarter or which ability a family dog lacks.
One cohort, two publications
This pair has a direct link: the 2026 follow-up includes 105 dogs from the earlier 111 first-test completers. Treat it as a follow-up, not two independent replications. The later reversal result adds exposure history, while uncontrolled interval practice and task structure still limit causal claims.
Selected vocabulary and repeated task exposure
The vocabulary study concerns selected Gifted Word Learners extending labels to new objects. The UBC follow-up concerns repeated exposure in a hand-touch sequence. Both include prior experience, but in different forms. The comparison cannot convert toy words or reversal change into a human age or an individual learning prognosis.
An editorial comparison of published designs, not a dog test, meta-analysis, validated research instrument, diagnosis or aptitude score. No animal information is entered, saved or sent. Sources did not review or endorse Taskwise.
