Blog 3: Addicted by Design - Why the Brain Loves the Shortcut

Scene 1: The Scroll That Never Ends

Picture this. It's midnight. You pick up your phone just to "check one notification." Suddenly, it's 2 AM. You've scrolled through hundreds of posts, reels, tweets - and your brain still whispers: "One more swipe."

Sounds familiar? That's not an accident. It's addiction by design.

From cigarettes to casinos, from TikTok to video games, the human brain is a playground where instant dopamine spikes win over rational long-term thinking. And if you peel back the curtain, you'll find an old friend underneath: the Markov Decision Process (MDP).

Scene 2: The Addiction MDP

Let's map addiction into our four-part MDP world.

• States (S):
  • Craving → Neutral → Satisfied (temporarily) → Withdrawal (back to craving).
• Actions (A):
  • Smoke, scroll, gamble, drink… OR resist.
• Rewards (R):
  • Immediate action → strong but short-lived dopamine reward.
  • Resistance → no immediate reward, but long-term health/social benefits.
• Transition Probabilities (P):
  • If you consume → high probability of looping back to craving (the cycle resets).
  • If you resist → probabilities shift slowly toward healthier states (but much harder, and slower to feel rewarding).

So the brain's environment gets "hacked": the reward function is skewed, and transitions push you back into the addictive loop.

Scene 3: The Short-Term Bias

Take smoking.

• Action: Smoke a cigarette now.
  • Reward = +10 (dopamine hit, stress relief).
  • Probability = 100% (always works instantly).
• Action: Resist smoking.
  • Reward = +0 immediately (sometimes even negative: stress, withdrawal symptoms).
  • Future Reward = +50 (better health, money saved, longer life).
  • Probability of feeling that reward now = ~0%.

Expected values look like this:

$$Q(\text{smoke}) = 10, \quad Q(\text{resist}) = 0 \; (\text{now})$$

Mathematically, resisting is the better long-term action. But because the brain discounts future rewards (\(\gamma\)) so heavily, the "+50" of health feels like "+0" in the moment.

So the rational mind loses. The shortcut wins.

Scene 4: Dopamine Hijack - The Brain's Cheat Code

Addiction essentially rewires the brain's reward system.

• Normally, dopamine is a signal for learning: "This action led to reward, maybe repeat it."
• In addiction, the brain stops learning and gets stuck in a feedback loop.

Every scroll, puff, or gamble reinforces the Q-value of the addictive action. The Q-learning update rule captures it beautifully:

$$Q(s,a) \leftarrow Q(s,a) + \alpha \Big(R + \gamma \max_{a'} Q(s',a') - Q(s,a)\Big)$$

Where:

• \(Q(s,a)\) = expected reward of action \(a\) in state \(s\).
• \(R\) = reward signal (dopamine spike).
• \(\gamma\) = discount factor.
• \(\alpha\) = learning rate.

In addictions, \(R\) is unnaturally high and immediate, so the brain keeps updating: "Scroll = good." "Smoke = good." "Spin the slot machine = good."

Scene 5: Casinos, Apps and The Science of Hooks

Here's where it gets dark: companies know this.

• Casinos use slot machines with variable rewards (sometimes win, mostly lose). The uncertainty boosts dopamine even more - the brain loves unpredictable jackpots.
• Social Media uses infinite scroll + likes as dopamine pellets. Each swipe is a new state, each like is a reward.
• Video Games use daily streaks, loot boxes, and leveling systems to exploit the same reinforcement cycle.

It's MDPs turned into billion-dollar industries, deliberately designing environments that trap human brains in addictive state-action loops.

Scene 6: The Psychology Angle

Why do people fall so easily into addiction?

• Short-term certainty beats long-term uncertainty. A cigarette always gives relief, but health is vague and delayed.
• Transition loops are sticky. Once you're in the craving–consumption–withdrawal cycle, the easiest path is to repeat.
• Future is discounted. The brain heavily downweights rewards that are not immediate.

Addiction is not just about weak willpower - it's about how our decision-making system (the MDP engine) gets hijacked.

Scene 7: Wrapping Up - The Trap and The Escape

The candy store and marshmallow test showed us the balance of now vs later. But addiction takes that balance and tilts it completely toward the present.

The result? A loop where rationality fades, and dopamine runs the show.

But here's the hopeful side: Therapies, habit re-engineering, and even AI-driven behavioral interventions can be seen as ways of reprogramming the MDP. By reshaping states, rewards, and transitions, we can guide the brain back toward healthier loops.

Next up in Blog 4, we'll step into the shadows of the mind - depression and anxiety. Here, the problem isn't too much dopamine, but too little perceived reward or too much perceived threat. We'll see how MDP helps us understand why people feel "stuck" in negative states, and how psychology works to shift their mental models.

Takeaway: Addiction isn't just a bad habit. It's the brain's decision-making engine - the MDP - running on hacked rewards and broken loops.