---
title: 'DR. WELL: Decentralized Neurosymbolic Planning'
url: https://www.emergentmind.com/topics/dr-well
type: topic
---

# DR. WELL: Decentralized Neurosymbolic Planning

DR. WELL most explicitly denotes **Dynamic Reasoning and Learning with Symbolic World Model for Embodied LLM-Based Multi-Agent Collaboration**, a **decentralized neurosymbolic framework** for embodied cooperative multi-agent planning in which agents negotiate task roles through structured communication, independently synthesize and execute symbolic plans, and improve over time using a shared **dynamic symbolic world model** that stores cross-episode task, plan, and outcome statistics [2511.04646]. It is designed for settings with **partial coordination**, **limited communication**, **asynchronous execution**, **task interdependence**, and **decentralized control**, and it replaces brittle trajectory-level synchronization with coordination through symbolic abstractions such as task identifiers, roles, and parameterized symbolic actions [2511.04646].

## 1. Conceptual basis and task model

DR. WELL is formulated for cooperative embodied tasks in which multiple agents must divide labor across interdependent subtasks. At time \(t\), the framework defines the set of agents ready to replan as
\[
\mathcal{A}_t^{r} = \{\, a_i \in \mathcal{A} \mid a_i \text{ is ready to replan at time } t \,\}, \qquad \mathcal{A}_t^{r} \subseteq \mathcal{A}.
\]
If \(|\mathcal{A}_t^{r}| > 0\), those agents enter a shared communication room for negotiation [2511.04646].

The main evaluation environment is **CUBE** (**Cooperative Push Block**), a grid-world block-pushing task. Blocks have size or weight \(w\), and pushing a block of weight \(w\) requires **\(w\) agents** to push simultaneously on the same face. Cooperation is therefore explicit in the task mechanics rather than treated as an emergent side effect [2511.04646].

The framework is motivated by the claim that direct trajectory-level coordination is fragile in embodied multi-agent settings. The cited reasons are that small timing errors can cascade into conflicts, raw LLM outputs are prompt-sensitive, decentralized agents cannot reliably align detailed action sequences without central control, and the number of agents and environmental conditions may vary [2511.04646]. DR. WELL addresses this by coordinating over symbolic plans rather than raw trajectories. This shifts the unit of synchronization from low-level motion to higher-level commitments that are reusable, synchronizable, and interpretable [2511.04646].

## 2. Negotiation protocol and role allocation

DR. WELL organizes coordination as a **two-phase negotiation protocol**: **negotiation / task allocation** followed by **symbolic planning and execution**. Negotiation occurs whenever one or more agents become idle and need a new commitment [2511.04646].

The participating agents are ordered in round-robin form,
\[
\sigma_t = (a_1,\ldots,a_m),
\]
and communicate through a shared buffer initialized with the current symbolic environment observation and a retrieved task-summary set from the world model [2511.04646]. For each task \(x\), the world model stores summary statistics
\[
r(x) = [\,\hat{\mu}_{x}^{\text{start}}, \hat{p}_{x}^{\text{succ}}, \hat{\mu}_{x}^{\text{duration}}, N_{x}^{\text{attempt}}\,],
\]
representing estimated mean start time, empirical success rate, mean duration, and number of attempts [2511.04646].

In the **proposal stage**, each agent proposes a candidate task and explains why:
\[
p_{a_j} = f_{\text{propose},a_j}\!\left(\phi_{a_j,t}, B_t^{(j-1)}\right),
\qquad
B_t^{(j)} = B_t^{(j-1)} \cup \{(a_j, p_{a_j})\}.
\]
A proposal includes a **task identifier** and a **natural-language rationale**. The rationale may mention spatial proximity, resource requirements, feasible tasks, and expected coordination needs [2511.04646].

After proposals, the system computes candidate tasks and retrieves the historically most effective team size for each:
\[
S_t = \{\,(\hat{n}(x),\, \hat{p}_{\text{succ}}(\hat{n}(x) \mid x )) \mid x \in X_t\,\},
\qquad
\hat{n}(x) = \arg\max_n \hat{p}_{\text{succ}}(n \mid x).
\]
Here, \(\hat{p}_{\text{succ}}(n \mid x)\) is the empirical probability that task \(x\) succeeds with team size \(n\), and \(\hat{n}(x)\) is the estimated optimal number of agents for that task [2511.04646].

In the **commitment stage**, each agent selects a task to commit to:
\[
c_{a_j} = f_{\text{commit},a_j}\!\left(\phi_{a_j,t}, B_t^{(m+j-1)}, S_t\right),
\qquad
B_t^{(m+j)} = B_t^{(m+j-1)} \cup \{(a_j, c_{a_j})\}.
\]
The resulting allocation is
\[
M_t = \{(a_j, c_{a_j}) \mid a_j \in \mathcal{A}_t^{r}\}, \qquad M_t : \mathcal{A}_t^{r} \rightarrow \mathcal{V}_{\text{task}}.
\]
Allocations are finalized only if they satisfy **consensus** and **quorum constraints**. A task requiring \(k\) agents proceeds only if at least \(k\) agents commit [2511.04646].

A defining design decision is communicative restraint. During negotiation, agents reveal only candidate tasks, reasoning, and final commitments; they do **not** reveal detailed symbolic plans [2511.04646]. This makes coordination depend on compatible commitments rather than on fragile full-plan sharing.

## 3. Symbolic planning and decentralized execution

After commitment, each agent plans and acts independently. The paper states that there is **no further inter-agent communication** during planning and execution. Each agent encodes its commitment into the LLM prompt, drafts a plan from scratch, refines that draft using the world model, executes symbolic actions through a controller, and replans when needed [2511.04646].

For agent \(a_j\), the initial plan draft is
\[
\pi_{a_j}^{\text{draft}} = f_{\text{draft},a_j}\!\left(\phi_{a_j,t}, M_t\right).
\]
This draft is a **plan prototype**: an argument-free symbolic action sequence that gives a high-level structural outline [2511.04646]. The world model then retrieves the top \(K\) historical plan prototypes for the committed task, ranked by success rate, and for each prototype the top \(L\) plan instances, ranked by success rate and then by shorter duration. The refined plan is
\[
\pi_{a_j}^{\text{inst}} = f_{\text{refine},a_j}\!\left(\pi_{a_j}^{\text{draft}}, R_{x_{a_j}}, \phi_{a_j,t}\right).
\]
This **self-refinement** mechanism reuses successful prototypes, avoids historically poor patterns, and specializes the current plan to the present context [2511.04646].

The symbolic action vocabulary is fixed, compact, and parameterized.

| Symbolic action | Parameters | Function |
|---|---|---|
| `WaitAgents` | `count = k`, `timeout = t` | Wait until \(k\) agents are idle, or until timeout |
| `Rendezvous` | `block_id`, `side`, `count = k`, `timeout = t` | Wait until \(k\) agents arrive at a specified side of a block, or timeout |
| `MoveToBlock` | `block_id`, `side` | Move and align to a specific side of the block |
| `Push` | `block_id`, `steps = n` | Push the block for \(n\) steps if aligned |
| `YieldFace` | `block_id`, `steps = n` | Move away from the block to free a face for collaborators |

Although the vocabulary is small, parameterization by block identifiers, sides, timeouts, and step counts yields a large effective planning space while preserving interpretability and synchronization [2511.04646].

Execution follows a **plan \(\rightarrow\) execute \(\rightarrow\) re-plan** cycle. The controller checks symbolic preconditions locally, translates symbolic actions into primitive moves, and advances execution like a state machine. Preconditions are checked locally by the controller, while postconditions and effects are confirmed by the environment [2511.04646]. If a cooperative action requires more aligned agents than are present, the controller places the agent in a waiting state and uses a timeout counter. If quorum is not reached before timeout, the action fails and the plan index advances. If the plan ends, the agent becomes idle and replanning starts [2511.04646]. This mechanism is explicitly intended to avoid deadlock while maintaining decentralized execution.

## 4. Dynamic symbolic world model

The shared world model is the framework’s central memory structure. It is defined as a **dynamic symbolic graph**
\[
\mathcal{G} = (V,E),
\]
with node sets and edges organized as
\[
\mathcal{V} = V_{\text{epi}} \cup V_{\text{task}} \cup V_{\text{proto}} \cup V_{\text{inst}},
\qquad
E = E_{\text{epi}\rightarrow\text{task}} \cup E_{\text{task}\rightarrow\text{proto}} \cup E_{\text{proto}\rightarrow\text{inst}}.
\]
The node types are episodes, tasks or blocks, plan prototypes, and plan instances; the edges connect episode to task, task to prototype, and prototype to instance [2511.04646].

Each instance node \(v \in V_{\text{inst}}\) is grounded to an observed outcome
\[
o(v) \in \{0,1\},
\]
where \(1\) denotes success and \(0\) failure. Upper-level nodes aggregate statistics from their descendants [2511.04646]. The world model is updated episodically. At update step \(k\), each episode contributes an additive subgraph
\[
\Delta \mathcal{G}_k = (\mathcal V_k^+, \mathcal E_k^+),
\]
and the graph evolves as
\[
\mathcal G_{k+1} = \mathcal G_k \cup \Delta \mathcal G_k.
\]
The graph therefore grows monotonically with experience [2511.04646].

The world model serves two explicit roles. As a **negotiation guidebook**, it supplies current timestep, number of active agents, historical completion rates, average durations, and estimated optimal team sizes. As a **plan library**, it supplies historical plan prototypes, detailed plan instances, success rates, average durations, number of attempts, and team-size information [2511.04646]. The framework also stores a full **world trace** including proposed tasks, commitments, symbolic actions executed, communication events, and completion outcomes [2511.04646]. This trace links symbolic intentions to concrete outcomes and supports later reasoning about what was attempted, what succeeded, what failed, how long it took, and how agents synchronized.

A plausible implication is that DR. WELL treats cross-episode learning not as end-to-end parameter adaptation alone, but as explicit accumulation of structured execution evidence. In the paper’s presentation, the world model is not an auxiliary explanation layer; it is the operational substrate that informs both allocation and refinement [2511.04646].

## 5. Empirical behavior in cooperative block-push tasks

The experimental environment is a grid world in which agents must push blocks into a goal zone with minimal steps. Observations are provided in two modalities: a multi-channel tensor and a symbolic state description. Symbolic observations include agent identifier and position, block size, block position, and distance to goal [2511.04646].

The main baseline is a **zero-shot symbolic planner** with a fixed prompt. It has **no negotiation**, **no task commitments**, **no communication**, **no plan revision**, and **no shared memory**. It repeatedly chooses the block closest to the goal and generates short plans independently [2511.04646]. The paper identifies a specific failure mode: multiple agents may redundantly work on the same easy block, while heavier blocks requiring cooperation remain unfinished [2511.04646].

Evaluation uses three metrics: **block completion outcome**, **completion time in wall-clock seconds**, and **completion time in environment steps** [2511.04646]. The reported findings are primarily qualitative and graphical rather than tabled as a single numeric benchmark. For the baseline, completion is inconsistent, many blocks remain unfinished—especially larger ones—time per episode stays nearly constant, and behavior shows no adaptation across episodes [2511.04646].

For DR. WELL, the reported trends are different. Almost all blocks are completed consistently after the early episodes; completion times decrease over episodes; environment steps also decline; task allocations become more stable after about **Episode 5**; overlap between agents decreases; and division of labor improves [2511.04646]. The framework incurs a trade-off: wall-clock time increases slightly because negotiation and replanning introduce overhead. The paper states that this overhead is offset by fewer wasted actions, better coordination, fewer failed attempts, faster completion in environment steps, and higher success rate [2511.04646].

The ablation-style analysis attributes the improvement to three components. **Negotiation** avoids all agents choosing the same task, matches task demands to team size, and enables consensus- and quorum-aware allocation. **Self-refinement** leverages historically successful prototypes, selects effective plan instances, and adapts them to the current context. The **dynamic world model** stores reusable symbolic patterns, accumulates success statistics, supports cross-episode reasoning, and captures evolving collaboration strategies [2511.04646]. The paper therefore argues that DR. WELL improves cooperative efficiency by combining structured communication at synchronization points, symbolic commitments instead of trajectory sharing, independent decentralized planning after role assignment, and execution-grounded shared memory [2511.04646].

## 6. Scope, limitations, and adjacent uses of the name

The framework is presented in a relatively structured setting. The paper identifies future extensions including reasoning about latent subgoals, partial local observations instead of full observability, interruption and re-negotiation when plans fail, richer in-group communication during sub-tasks, more realistic dynamic task allocation, the ability to redirect or interrupt teammates, and probabilistic symbolic outcomes for uncertainty [2511.04646]. A common misconception would be to treat DR. WELL as a general solution to all multi-agent embodiment settings; the stated future directions indicate that the current formulation assumes an environment well suited to explicit symbolic abstractions [2511.04646].

The name also has adjacent meanings in health-oriented AI discourse. One paper states that, in its context, **“DR. WELL”** refers to the same idea as a **Dr. Watson-type AI**: not an autonomous oracle, but a supportive, question-asking, contradiction-finding assistant that helps the human expert think better and decide better [2106.13322]. In that formulation, a Dr. Watson-type system **does not offer a specific solution**, but instead **analyzes the user’s solution** to reveal logical inconsistencies or insufficient data, using **game situations**, **psychological techniques**, and **visual presentation of information** to stimulate the user’s thinking [2106.13322]. Another paper on Reddit-based mental health analysis states that its wellness concept extraction framework would be especially valuable for a wellness-oriented assistant like **“Dr. Well”**, because it moves from generic risk detection toward interpretable, dimension-specific support [2308.13710].

These usages do not describe the same technical system. The embodied multi-agent DR. WELL framework concerns decentralized neurosymbolic planning with symbolic commitments and a dynamic world model [2511.04646], whereas the adjacent “Dr. Well” usages concern wellness-oriented or Dr. Watson-type human-support systems [2106.13322; 2308.13710]. The shared motif is not architecture but assistance through structured, interpretable reasoning.

Source: https://www.emergentmind.com/topics/dr-well