Pith. sign in

REVIEW 7 cited by

Core Knowledge Deficits in Multi-Modal Language Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.10855 v4 pith:CMLQZHZF submitted 2024-10-06 cs.CL cs.AIcs.CV

Core Knowledge Deficits in Multi-Modal Language Models

classification cs.CL cs.AIcs.CV
keywords coreknowledgemllmsabilitiesmodelscognitivedeficitshigh-level
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

While Multi-modal Large Language Models (MLLMs) demonstrate impressive abilities over high-level perception and reasoning, their robustness in the wild remains limited, often falling short on tasks that are intuitive and effortless for humans. We examine the hypothesis that these deficiencies stem from the absence of core knowledge--rudimentary cognitive abilities innate to humans from early childhood. To explore the core knowledge representation in MLLMs, we introduce CoreCognition, a large-scale benchmark encompassing 12 core knowledge concepts grounded in developmental cognitive science. We evaluate 230 models with 11 different prompts, leading to a total of 2,530 data points for analysis. Our experiments uncover four key findings, collectively demonstrating core knowledge deficits in MLLMs: they consistently underperform and show reduced, or even absent, scalability on low-level abilities relative to high-level ones. Finally, we propose Concept Hacking, a novel controlled evaluation method that reveals MLLMs fail to progress toward genuine core knowledge understanding, but instead rely on shortcut learning as they scale.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. VisualFLIP: Do Predictions Depend on Task-Critical Visual Evidence in Multimodal Reasoning?

    cs.CV 2026-06 unverdicted novelty 7.0

    A paired-image benchmark reveals that many MLLMs fail to update predictions when task-critical visual evidence changes, even when they answer individual images correctly.

  2. LEVANTE-bench: Multi-Scale Comparison of VLMs to Children Using Cognitive Tasks (or, "Is Your VLM Smarter Than a 5th Grader?")

    cs.LG 2026-06 unverdicted novelty 7.0

    VLMs show partial alignment with children's performance on six cognitive tasks, with stronger models matching better at task and item levels but struggling on matrix reasoning and mental rotation.

  3. SpatialAct: Probing Spatial Reasoning-to-Action Capabilities of VLM Agents in 3D Scenes

    cs.CV 2026-05 unverdicted novelty 7.0

    SpatialAct benchmark shows VLMs handle isolated spatial reasoning but fail to maintain coherent spatial beliefs and produce reliable actions in multi-turn 3D interactions, underperforming humans.

  4. Vision Language Models Cannot Reason About Physical Transformation

    cs.AI 2026-03 accept novelty 6.5

    Current VLMs cannot maintain transformation-invariant representations of number, length, volume or size and instead rely on textual invariance priors that reverse on matched non-conserving controls.

  5. VKnowU: Evaluating Visual Knowledge Understanding in Multimodal LLMs

    cs.CV 2025-11 conditional novelty 6.0

    A 1,680-question video benchmark shows leading multimodal models lag humans by ~15 points on visual knowledge, and a See-Think-Answer RL-trained model narrows the gap.

  6. The Landscape of Agentic Reinforcement Learning for LLMs: A Survey

    cs.AI 2025-09 accept novelty 6.0

    Survey that defines agentic RL for LLMs via POMDPs, introduces a taxonomy of planning/tool-use/memory/reasoning capabilities and domains, and compiles open environments from over 500 papers.

  7. ThinkJEPA: Empowering Latent World Models with Large Vision-Language Reasoning Model

    cs.CV 2026-03 conditional novelty 5.5

    Dual-temporal VLM guidance injected into a JEPA predictor via multi-layer pyramid features improves hand-manipulation trajectory forecasting over VLM-only and JEPA-only baselines.