Hierarchy of LLM Reasoning Techniques

Hierarchy of LLM Reasoning Techniques

This outlines how these prompting methods generally relate, building from the simplest approach:

1. Standard Prompting (Baseline)
   • Description: Asks for a direct answer without requesting intermediate steps.
   • Process: Input -> Output

2. Chain of Thought (CoT) Prompting (Foundational Step-by-Step Reasoning) [1]
   • Description: Elicits a single, linear sequence of reasoning steps (text-based) before providing the final answer. Improves accuracy on complex tasks by breaking them down.
   • Process: Input -> Step 1 -> Step 2 -> ... -> Output
   • Sub-types:
     • Zero-Shot CoT: Uses a trigger phrase like “Let’s think step by step” without examples [5].
     • Few-Shot CoT: Provides examples of step-by-step reasoning within the prompt [6].

3. Domain-Specific Adaptations of CoT
   • Description: Applies the step-by-step reasoning concept to specific data types or structures.
   • Chain of Table (CoTbl): Adapts CoT specifically for structured tabular data. Reasoning steps involve generating and applying table operations (like filter, group by, sort) iteratively, producing intermediate tables [2, 7].
     • Process: Input (Query + Table_0) -> Operation_1 -> Table_1 -> Operation_2 -> Table_2 -> ... -> Output

4. Enhancements & Generalizations of CoT
   • Description: These techniques build upon, modify, or extend the basic CoT structure to improve robustness or handle more complex problems.
   • Self-Consistency (SC): An ensemble method applied on top of CoT [3].
     • Process:
       1. Generate multiple independent CoT reasoning paths for the same input (often using sampling).
       2. Aggregate the final answers from each path.
       3. Select the most frequent answer (majority vote) as the final, more reliable output.
     • Relationship to CoT: Uses CoT as a base generator but adds a consensus layer for robustness.
   • Tree of Thoughts (ToT): A generalization of CoT enabling non-linear, exploratory reasoning [4].
     • Process:
       1. Generates multiple possible thoughts or reasoning steps (branches) at each stage.
       2. Evaluates the potential of different branches towards reaching the solution.
       3. Uses search algorithms (like BFS or DFS) to explore the “thought tree,” allowing for strategic planning and backtracking from unpromising paths.
     • Relationship to CoT: Extends CoT’s linear path into a branching structure for complex exploration and decision-making.

Taxonomy Summary:

• CoT is the core technique establishing linear, step-by-step textual reasoning.
• CoTbl is a specialized version of CoT tailored for structured operations on tables.
• SC is an enhancement layer applied over CoT to increase answer reliability through consensus.
• ToT is a more complex generalization of CoT, moving from linear chains to branching trees for exploration and strategic problem-solving.

This hierarchy shows a progression from the simplest form of prompting to increasingly sophisticated methods designed to handle specific data types, improve robustness, or tackle more complex, non-linear problems.

---

Reference List

[1] NVIDIA. (n.d.). What is Chain of Thought (CoT) Prompting? NVIDIA. Retrieved October 20, 2025, from https://www.nvidia.com/en-us/glossary/cot-prompting/

[2] Google Research. (2024, March 11). Chain-of-table: Evolving tables in the reasoning chain for table understanding. Google Research Blog. Retrieved October 20, 2025, from https://research.google/blog/chain-of-table-evolving-tables-in-the-reasoning-chain-for-table-understanding/

[3] PromptHub. (n.d.). Self-Consistency and Universal Self-Consistency Prompting. Retrieved October 20, 2025, from https://www.prompthub.us/blog/self-consistency-and-universal-self-consistency-prompting

[4] Emeritus. (2024, January 22). Tree of Thoughts Prompting; How Does it Enhance AI Results? Retrieved October 20, 2025, from https://emeritus.org/blog/tree-of-thoughts-prompting/

[5] IBM. (n.d.). What is chain of thought (CoT) prompting? IBM. Retrieved October 20, 2025, from https://www.ibm.com/think/topics/chain-of-thoughts

[6] Codecademy. (n.d.). Chain of Thought Prompting Explained (with examples). Retrieved October 20, 2025, from https://www.codecademy.com/article/chain-of-thought-cot-prompting

[7] Relevance AI. (n.d.). Master Chain-of-Table Prompting for Effective Data Analysis. Retrieved October 20, 2025, from https://relevanceai.com/prompt-engineering/master-chain-of-table-prompting-for-effective-data-analysis