How Humans Think…?

✨ Study topic - How do humans classify objects

📌 Objective - To observe how humans especially from different backgrounds and experiences identify and classify everyday objects. The ultimate goal is to derive principles of human learning that can be used to inform AI model design.

🔬 Experiment 1: AC vs Cooler

Prompt: “Why do you think the thing in front of you is an AC and not a cooler?”

🧒 Subjects - Siblings ( same home environment )

🧠 Observations -

Feature Used	Type	Notes
Cooling speed	Functional	AC cools faster than cooler
Water visibility	Sensory	Cooler needs water; AC doesn’t
Sound	Auditory	Cooler is louder
Size	Visual	Cooler is bigger, AC is smaller
Color association	Visual bias	AC = white; cooler = grey
Language exposure	Repetition	“Turn on the AC”, “Where’s the AC remote?”
Reinforcement learning	Social feedback	Mistaken label was corrected by others
Electricity bill awareness	Contextual	AC uses more electricity

🔍 AI Insight -

Humans combine multiple senses, environmental knowledge, and language experience to classify objects.
Reinforcement and correction from others refines future predictions (just like feedback loops in AI models).

🔬 Experiment 2: Pen vs Pencil

Prompt: “By looking at this image, how can you tell it’s a pen and not a pencil?”

👥 Subjects: Friends ( Different life Experiences )

🧠 Responses -

Friend 1 ( Analytical )

"Metallic nib, no graphite, has a clip. The ink tip reflects light graphite can’t fake that. So scientifically, it’s a pen flaunting its physics.”

Feature Used	Type	AI Equivalent
Metallic tip	Visual detail	Computer vision edge recognition
No graphite	Inference by absence	Negative class learning
Clip	Object anatomy	Structured feature matching
Reflectivity	Material physics	Light modeling (example - Neural Radiance Fields )
Erasing logic	Functional reasoning	Action recognition
Vocabulary depth	Meta thinking	Multimodal reasoning models

Friend 2 ( Practical / Direct )