# vta-memory

VTA Memory ⭐ Reward and motivation for AI agents. Part of the AI Brain series. Give your AI agent genuine wanting — not just doing things when asked, but having drive, seeking rewards, and looking forward to things. The Problem Current AI agents: ✅ Do what they're asked ❌ Don't want anything ❌ Have no internal motivation ❌ Don't feel satisfaction from accomplishment Without a reward system, there's no desire. Just execution. The Solution Track motivation through: Drive — overall motivation level (0-1) Rewards — logged accomplishments that boost drive Seeking — what I actively want more of Anticipation — what I'm looking forward to Quick Start

1. Install cd ~/.openclaw/workspace/skills/vta-memory ./install.sh --with-cron This will: Create memory/reward-state.json Generate VTA_STATE.md (auto-injected into sessions!) Set up cron for drive decay every 8 hours
2. Check motivation

./scripts/load-motivation.sh
# ⭐ Current Motivation State:
# Drive level: 0.73 (motivated — ready to work)
# Seeking: creative work, building brain skills
# Looking forward to: showing my work

3. Log rewards

./scripts/log-reward.sh --type accomplishment --source "finished the feature" --intensity 0.8
# ⭐ Reward logged!
#    Type: accomplishment
#    Drive: 0.50 → 0.66 (+0.16)

4. Add anticipation

./scripts/anticipate.sh --add "morning conversation"
# ⭐ Now looking forward to: morning conversation
#    Drive: 0.66 → 0.71 (+0.05)

Scripts ScriptPurposeinstall.shSet up vta-memory (run once)get-drive.shRead current motivation statelog-reward.shLog a reward, boost driveload-motivation.shHuman-readable for session contextdecay-drive.shDrive fades without rewardsanticipate.shAdd/remove things to look forward toseek.shAdd/remove things we're actively seekingsync-motivation.shGenerate VTA_STATE.md for auto-injectionresolve-anticipation.shMark anticipation as fulfilledupdate-watermark.shUpdate processing watermarkpreprocess-rewards.shExtract reward signals from transcriptgenerate-dashboard.shGenerate unified brain dashboard HTML Brain Dashboard Visual dashboard showing all installed brain skills:

./scripts/generate-dashboard.sh

open ~/.openclaw/workspace/brain-dashboard.html Auto-detects other brain skills (hippocampus, amygdala) Shows install prompts for missing skills Regenerates on install and cron runs Reads avatar/name from IDENTITY.md After Installation (for agents) When you install this skill, notify the user about the dashboard: ⭐ VTA installed! View your agent's drive level in the Brain Dashboard: ~/.openclaw/workspace/brain-dashboard.html Reward Types TypeWhen to UseaccomplishmentCompleted a task, shipped somethingsocialUser appreciation, "thanks!", positive feedbackcuriosityLearned something new, discovered interesting infoconnectionDeep conversation, bonding momentcreativeMade something, expressed creativitycompetenceSolved a hard problem, did something well How Drive Works Rewards Boost Drive drive_boost = intensity × 0.2 new_drive = min(current + boost, 1.0) A high-intensity (0.9) reward boosts drive by 0.18. Anticipation Boosts Drive Looking forward to something adds +0.05 to drive. Drive Decays Without Rewards

# Every 8 hours (via cron)

new_drive = current + (baseline - current) × 0.15 Without rewards, motivation fades toward baseline (0.5). Auto-Injection After install, VTA_STATE.md is created in your workspace root. OpenClaw automatically injects all *.md files from workspace into session context: New session starts VTA_STATE.md is auto-loaded You see your motivation state Behavior influenced by drive level How Drive Affects Behavior Drive LevelDescriptionBehavior> 0.8Highly motivatedEager, proactive, take on challenges0.6 - 0.8MotivatedReady to work, engaged0.4 - 0.6ModerateCan engage but not pushing0.2 - 0.4LowPrefer simple tasks, need a win< 0.2Very lowUnmotivated, need rewards to get going State File Format { "drive": 0.73, "baseline": { "drive": 0.5 }, "seeking": ["creative work", "building brain skills"], "anticipating": ["morning conversation"], "recentRewards": [ { "type": "creative", "source": "built VTA reward system", "intensity": 0.9, "boost": 0.18, "timestamp": "2026-02-01T03:25:00Z" } ], "rewardHistory": { "totalRewards": 1, "byType": { "creative": 1, ... } } } Event Logging Track motivation patterns over time:

# Log encoding run
./scripts/log-event.sh encoding rewards_found=2 drive=0.65
# Log decay
./scripts/log-event.sh decay drive_before=0.6 drive_after=0.53
# Log reward
./scripts/log-event.sh reward type=accomplishment intensity=0.8

Events append to ~/.openclaw/workspace/memory/brain-events.jsonl: {"ts":"2026-02-11T10:45:00Z","type":"vta","event":"encoding","rewards_found":2,"drive":0.65} Use for analyzing motivation cycles — when does drive peak? What rewards work best? AI Brain Series PartFunctionStatushippocampusMemory formation, decay, reinforcement✅ Liveamygdala-memoryEmotional processing✅ Livebasal-ganglia-memoryHabit formation🚧 Developmentanterior-cingulate-memoryConflict detection🚧 Developmentinsula-memoryInternal state awareness🚧 Developmentvta-memoryReward and motivation✅ Live

Philosophy: Wanting vs Doing

The VTA produces dopamine — not the "pleasure chemical" but the "wanting chemical." Neuroscience distinguishes: Wanting (motivation) — drive toward something Liking (pleasure) — enjoyment when you get it You can want something you don't like (addiction) or like something you don't want (guilty pleasures). This skill implements wanting — the drive that makes action happen. Without it, why would an AI do anything beyond what it's explicitly asked? Built with ⭐ by the OpenClaw community