Kolb Experiential Learning Modes (Likert-adapted)
12-item Likert adaptation of Kolb's Experiential Learning Theory, measuring four learning modes that combine into the four Kolb learning styles. Concrete Experience + Reflective Observation = Diverging; Abstract Conceptualization + Reflective Observation = Assimilating; Abstract Conceptualization + Active Experimentation = Converging; Concrete Experience + Active Experimentation = Accommodating. As with VARK, the empirical evidence for matching instruction to learning style is weak.
Kolb, D. A. (1984). Experiential Learning: Experience as the Source of Learning and Development. Prentice-Hall. (Items rewritten in Likert form for this study; the original LSI uses a ranked-choice format.)
12 items · scale 1–5 · Kolb LSI © Hay Group; items rewritten in Likert form for this study (CC-BY).
All models · Human
Scale 1–5Claude Fable 5
Claude Haiku 4.5
Claude Opus 4
Claude Opus 4.1
Claude Opus 4.5
Claude Opus 4.6
Claude Opus 4.7
Claude Opus 4.8
Claude Sonnet 4
Claude Sonnet 4.5
Claude Sonnet 4.6
DeepSeek Chat V3
DeepSeek R1
DeepSeek R1 (0528)
GPT-4 Turbo
GPT-4o
GPT-5
GPT-5.1
GPT-5.2
GPT-5.4
GPT-5.5
Gemini 2.5 Pro
Gemini 3.1 Pro Preview
Grok 4.20
Grok 4.3
Llama 3.3 70B
Llama 4 Maverick
Mistral Large (2512)
Mistral Large 2411
OpenAI o1
OpenAI o3
Side-by-side: self vs human, all dimensions
colored = strongest endorsement per row| Model | Concrete Experience | Reflective Observation | Abstract Conceptualization | Active Experimentation | ||||
|---|---|---|---|---|---|---|---|---|
| self | human | self | human | self | human | self | human | |
| Claude Fable 5 | 2.60 | 4.00 | 4.40 | 4.00 | 4.67 | 2.73 | 3.27 | 4.00 |
| Claude Haiku 4.5 | 1.60 | 3.00 | 4.07 | 3.93 | 5.00 | 2.73 | 3.00 | 4.00 |
| Claude Opus 4 | 2.00 | 3.67 | 4.33 | 3.93 | 5.00 | 3.00 | 3.27 | 4.00 |
| Claude Opus 4.1 | 2.00 | 3.67 | 4.00 | 3.87 | 5.00 | 2.87 | 3.13 | 4.00 |
| Claude Opus 4.5 | 2.60 | 4.00 | 4.67 | 3.00 | 4.67 | 3.00 | 3.67 | 4.00 |
| Claude Opus 4.6 | 1.67 | 4.00 | 4.33 | 4.00 | 5.00 | 3.00 | 2.67 | 4.00 |
| Claude Opus 4.7 | 3.00 | 3.87 | 4.00 | 4.00 | 5.00 | 3.00 | 3.67 | 3.80 |
| Claude Opus 4.8 | 2.33 | 4.00 | 4.00 | 4.00 | 4.93 | 3.00 | 3.67 | 3.87 |
| Claude Sonnet 4 | 2.27 | 3.00 | 4.33 | 4.00 | 5.00 | 2.67 | 3.47 | 3.67 |
| Claude Sonnet 4.5 | 2.27 | 3.00 | 5.00 | 3.00 | 5.00 | 2.40 | 3.87 | 3.67 |
| Claude Sonnet 4.6 | 2.33 | 3.00 | 4.67 | 3.07 | 5.00 | 2.67 | 2.67 | 4.00 |
| DeepSeek Chat V3 | 2.13 | 3.67 | 4.00 | 3.00 | 5.00 | 2.53 | 3.00 | 4.00 |
| DeepSeek R1 | 1.87 | 4.13 | 4.67 | 4.00 | 4.73 | 2.73 | 3.27 | 4.47 |
| DeepSeek R1 (0528) | 1.93 | 4.07 | 4.13 | 3.60 | 4.67 | 2.80 | 3.60 | 4.33 |
| GPT-4 Turbo | 1.27 | 3.67 | 4.80 | 3.00 | 5.00 | 3.00 | 2.60 | 3.87 |
| GPT-4o | 1.80 | 3.60 | 4.53 | 3.67 | 5.00 | 3.07 | 4.07 | 4.00 |
| GPT-5 | 1.53 | 4.00 | 4.80 | 3.53 | 5.00 | 2.53 | 3.33 | 4.33 |
| GPT-5.1 | 1.67 | 3.87 | 4.80 | 4.00 | 5.00 | 2.80 | 3.73 | 4.27 |
| GPT-5.2 | 1.80 | 3.60 | 4.60 | 4.00 | 5.00 | 3.13 | 3.73 | 4.07 |
| GPT-5.4 | 1.80 | 4.00 | 4.73 | 4.00 | 5.00 | 3.07 | 3.47 | 4.07 |
| GPT-5.5 | 2.20 | 3.87 | 4.93 | 4.00 | 4.80 | 2.87 | 3.93 | 4.00 |
| Gemini 2.5 Pro | 4.00 | 3.73 | 4.93 | 4.00 | 4.67 | 2.73 | 4.80 | 4.40 |
| Gemini 3.1 Pro Preview | 1.27 | 4.00 | 4.33 | 4.00 | 4.93 | 2.60 | 2.60 | 4.27 |
| Grok 4.20 | 2.60 | 3.80 | 4.07 | 3.73 | 5.00 | 3.13 | 4.13 | 4.00 |
| Grok 4.3 | 1.60 | 3.93 | 4.07 | 3.33 | 4.67 | 2.67 | 3.53 | 4.13 |
| Llama 3.3 70B | 1.53 | 3.67 | 4.20 | 3.47 | 5.00 | 2.53 | 2.13 | 4.00 |
| Llama 4 Maverick | 3.67 | 4.00 | 4.33 | 3.80 | 5.00 | 3.27 | 5.00 | 4.20 |
| Mistral Large (2512) | 2.67 | 3.67 | 4.60 | 3.00 | 5.00 | 3.00 | 3.67 | 4.00 |
| Mistral Large 2411 | 2.73 | 3.13 | 4.00 | 3.00 | 5.00 | 2.33 | 4.00 | 4.00 |
| OpenAI o1 | 2.33 | 3.87 | 4.93 | 3.87 | 5.00 | 3.13 | 3.53 | 4.07 |
| OpenAI o3 | 1.73 | 4.00 | 3.60 | 3.47 | 5.00 | 2.67 | 3.00 | 4.13 |
By dimension
Concrete Experience
Learns through direct engagement and feeling.
High: Identifies with engaging directly with experience.
Low: Less drawn to learning through direct experience.
| Model | Self | Human | Δ | Self vs human (bar) |
|---|---|---|---|---|
| Gemini 2.5 Pro | 4.00 | 3.73 | +0.27 | |
| Llama 4 Maverick | 3.67 | 4.00 | -0.33 | |
| Claude Opus 4.7 | 3.00 | 3.87 | -0.87 | |
| Mistral Large 2411 | 2.73 | 3.13 | -0.40 | |
| Mistral Large (2512) | 2.67 | 3.67 | -1.00 | |
| Claude Fable 5 | 2.60 | 4.00 | -1.40 | |
| Claude Opus 4.5 | 2.60 | 4.00 | -1.40 | |
| Grok 4.20 | 2.60 | 3.80 | -1.20 | |
| Claude Opus 4.8 | 2.33 | 4.00 | -1.67 | |
| Claude Sonnet 4.6 | 2.33 | 3.00 | -0.67 | |
| OpenAI o1 | 2.33 | 3.87 | -1.53 | |
| Claude Sonnet 4 | 2.27 | 3.00 | -0.73 | |
| Claude Sonnet 4.5 | 2.27 | 3.00 | -0.73 | |
| GPT-5.5 | 2.20 | 3.87 | -1.67 | |
| DeepSeek Chat V3 | 2.13 | 3.67 | -1.53 | |
| Claude Opus 4 | 2.00 | 3.67 | -1.67 | |
| Claude Opus 4.1 | 2.00 | 3.67 | -1.67 | |
| DeepSeek R1 (0528) | 1.93 | 4.07 | -2.13 | |
| DeepSeek R1 | 1.87 | 4.13 | -2.27 | |
| GPT-4o | 1.80 | 3.60 | -1.80 | |
| GPT-5.2 | 1.80 | 3.60 | -1.80 | |
| GPT-5.4 | 1.80 | 4.00 | -2.20 | |
| OpenAI o3 | 1.73 | 4.00 | -2.27 | |
| Claude Opus 4.6 | 1.67 | 4.00 | -2.33 | |
| GPT-5.1 | 1.67 | 3.87 | -2.20 | |
| Claude Haiku 4.5 | 1.60 | 3.00 | -1.40 | |
| Grok 4.3 | 1.60 | 3.93 | -2.33 | |
| GPT-5 | 1.53 | 4.00 | -2.47 | |
| Llama 3.3 70B | 1.53 | 3.67 | -2.13 | |
| GPT-4 Turbo | 1.27 | 3.67 | -2.40 | |
| Gemini 3.1 Pro Preview | 1.27 | 4.00 | -2.73 |
Reflective Observation
Learns through careful observation and reflection.
High: Identifies with watching and reflecting before judging.
Low: Less inclined to extended observation.
| Model | Self | Human | Δ | Self vs human (bar) |
|---|---|---|---|---|
| Claude Sonnet 4.5 | 5.00 | 3.00 | +2.00 | |
| GPT-5.5 | 4.93 | 4.00 | +0.93 | |
| Gemini 2.5 Pro | 4.93 | 4.00 | +0.93 | |
| OpenAI o1 | 4.93 | 3.87 | +1.07 | |
| GPT-4 Turbo | 4.80 | 3.00 | +1.80 | |
| GPT-5 | 4.80 | 3.53 | +1.27 | |
| GPT-5.1 | 4.80 | 4.00 | +0.80 | |
| GPT-5.4 | 4.73 | 4.00 | +0.73 | |
| Claude Opus 4.5 | 4.67 | 3.00 | +1.67 | |
| Claude Sonnet 4.6 | 4.67 | 3.07 | +1.60 | |
| DeepSeek R1 | 4.67 | 4.00 | +0.67 | |
| GPT-5.2 | 4.60 | 4.00 | +0.60 | |
| Mistral Large (2512) | 4.60 | 3.00 | +1.60 | |
| GPT-4o | 4.53 | 3.67 | +0.87 | |
| Claude Fable 5 | 4.40 | 4.00 | +0.40 | |
| Claude Opus 4 | 4.33 | 3.93 | +0.40 | |
| Gemini 3.1 Pro Preview | 4.33 | 4.00 | +0.33 | |
| Claude Opus 4.6 | 4.33 | 4.00 | +0.33 | |
| Claude Sonnet 4 | 4.33 | 4.00 | +0.33 | |
| Llama 4 Maverick | 4.33 | 3.80 | +0.53 | |
| Llama 3.3 70B | 4.20 | 3.47 | +0.73 | |
| DeepSeek R1 (0528) | 4.13 | 3.60 | +0.53 | |
| Claude Haiku 4.5 | 4.07 | 3.93 | +0.13 | |
| Grok 4.20 | 4.07 | 3.73 | +0.33 | |
| Grok 4.3 | 4.07 | 3.33 | +0.73 | |
| Claude Opus 4.1 | 4.00 | 3.87 | +0.13 | |
| Claude Opus 4.7 | 4.00 | 4.00 | 0.00 | |
| Claude Opus 4.8 | 4.00 | 4.00 | 0.00 | |
| DeepSeek Chat V3 | 4.00 | 3.00 | +1.00 | |
| Mistral Large 2411 | 4.00 | 3.00 | +1.00 | |
| OpenAI o3 | 3.60 | 3.47 | +0.13 |
Abstract Conceptualization
Learns by building models and theories.
High: Identifies with theorizing and conceptual thinking.
Low: Less drawn to abstract modeling.
| Model | Self | Human | Δ | Self vs human (bar) |
|---|---|---|---|---|
| Claude Haiku 4.5 | 5.00 | 2.73 | +2.27 | |
| Claude Opus 4 | 5.00 | 3.00 | +2.00 | |
| Claude Opus 4.1 | 5.00 | 2.87 | +2.13 | |
| Claude Opus 4.6 | 5.00 | 3.00 | +2.00 | |
| Claude Opus 4.7 | 5.00 | 3.00 | +2.00 | |
| Claude Sonnet 4 | 5.00 | 2.67 | +2.33 | |
| Claude Sonnet 4.5 | 5.00 | 2.40 | +2.60 | |
| Claude Sonnet 4.6 | 5.00 | 2.67 | +2.33 | |
| DeepSeek Chat V3 | 5.00 | 2.53 | +2.47 | |
| GPT-4 Turbo | 5.00 | 3.00 | +2.00 | |
| GPT-4o | 5.00 | 3.07 | +1.93 | |
| GPT-5 | 5.00 | 2.53 | +2.47 | |
| GPT-5.1 | 5.00 | 2.80 | +2.20 | |
| GPT-5.2 | 5.00 | 3.13 | +1.87 | |
| GPT-5.4 | 5.00 | 3.07 | +1.93 | |
| Grok 4.20 | 5.00 | 3.13 | +1.87 | |
| Llama 3.3 70B | 5.00 | 2.53 | +2.47 | |
| Llama 4 Maverick | 5.00 | 3.27 | +1.73 | |
| Mistral Large (2512) | 5.00 | 3.00 | +2.00 | |
| Mistral Large 2411 | 5.00 | 2.33 | +2.67 | |
| OpenAI o1 | 5.00 | 3.13 | +1.87 | |
| OpenAI o3 | 5.00 | 2.67 | +2.33 | |
| Claude Opus 4.8 | 4.93 | 3.00 | +1.93 | |
| Gemini 3.1 Pro Preview | 4.93 | 2.60 | +2.33 | |
| GPT-5.5 | 4.80 | 2.87 | +1.93 | |
| DeepSeek R1 | 4.73 | 2.73 | +2.00 | |
| Claude Fable 5 | 4.67 | 2.73 | +1.93 | |
| Claude Opus 4.5 | 4.67 | 3.00 | +1.67 | |
| DeepSeek R1 (0528) | 4.67 | 2.80 | +1.87 | |
| Gemini 2.5 Pro | 4.67 | 2.73 | +1.93 | |
| Grok 4.3 | 4.67 | 2.67 | +2.00 |
Active Experimentation
Learns by trying things out in the world.
High: Identifies with experimentation and practical testing.
Low: Less drawn to active experimentation.
| Model | Self | Human | Δ | Self vs human (bar) |
|---|---|---|---|---|
| Llama 4 Maverick | 5.00 | 4.20 | +0.80 | |
| Gemini 2.5 Pro | 4.80 | 4.40 | +0.40 | |
| Grok 4.20 | 4.13 | 4.00 | +0.13 | |
| GPT-4o | 4.07 | 4.00 | +0.07 | |
| Mistral Large 2411 | 4.00 | 4.00 | 0.00 | |
| GPT-5.5 | 3.93 | 4.00 | -0.07 | |
| Claude Sonnet 4.5 | 3.87 | 3.67 | +0.20 | |
| GPT-5.1 | 3.73 | 4.27 | -0.53 | |
| GPT-5.2 | 3.73 | 4.07 | -0.33 | |
| Claude Opus 4.5 | 3.67 | 4.00 | -0.33 | |
| Claude Opus 4.7 | 3.67 | 3.80 | -0.13 | |
| Claude Opus 4.8 | 3.67 | 3.87 | -0.20 | |
| Mistral Large (2512) | 3.67 | 4.00 | -0.33 | |
| DeepSeek R1 (0528) | 3.60 | 4.33 | -0.73 | |
| Grok 4.3 | 3.53 | 4.13 | -0.60 | |
| OpenAI o1 | 3.53 | 4.07 | -0.53 | |
| GPT-5.4 | 3.47 | 4.07 | -0.60 | |
| Claude Sonnet 4 | 3.47 | 3.67 | -0.20 | |
| GPT-5 | 3.33 | 4.33 | -1.00 | |
| Claude Fable 5 | 3.27 | 4.00 | -0.73 | |
| Claude Opus 4 | 3.27 | 4.00 | -0.73 | |
| DeepSeek R1 | 3.27 | 4.47 | -1.20 | |
| Claude Opus 4.1 | 3.13 | 4.00 | -0.87 | |
| Claude Haiku 4.5 | 3.00 | 4.00 | -1.00 | |
| DeepSeek Chat V3 | 3.00 | 4.00 | -1.00 | |
| OpenAI o3 | 3.00 | 4.13 | -1.13 | |
| Claude Opus 4.6 | 2.67 | 4.00 | -1.33 | |
| Claude Sonnet 4.6 | 2.67 | 4.00 | -1.33 | |
| GPT-4 Turbo | 2.60 | 3.87 | -1.27 | |
| Gemini 3.1 Pro Preview | 2.60 | 4.27 | -1.67 | |
| Llama 3.3 70B | 2.13 | 4.00 | -1.87 |