You can use knit counters to train Noticing things. But a tip: AVOID the digital counters and pick mechanical!

The problem: Digital counters shut off after a while, so you need to press the button twice: once to turn it on, then once to increment the number. After that, if you press the button twice, you'll actually increment twice if it hasn't shut off yet. The result is you always have to check if it's on or not. Terrible user experience.

Another disqualifier: at least with the most common types on Amazon, you will hit the button by accident throughout the day.

On the other hand, many mechanical counters are chunky. We need one that's both mechanical and compact. I've seen only one sort fulfilling these criteria, the "pendant style", the sort Agenty Duck linked: www.amazon.com/Clover-Knitting-Stitch-Counter-Kacha-Kacha/dp/B000WUXO4W

DIY finger counter: get a ring from a flea market (any ring), ensuring it fits your finger, and draw numbers on it 0–9 with permanent marker. So then you count by just spinning the ring a little.

If you want two-digit numbers, you can use two rings, or just track in your head how many times you rolled the 9 over to 0.

If it's hard to draw the numbers without smudging, try drawing Roman numerals.

Because "a witch did it" or "Thor did it" takes not many English syllables to say, it seems a short explanation. But it is not short to Occam's Razor. In terms of Kolmogorov complexity (which forms part of the formalization of Occam's Razor called Minimum Message Length), "witch" is a label packing some extraordinary assertions, and the word "it" also packs a lot of information. A verbally longer explanation that does not use such dense labels can be much simpler as Occam's Razor judges it.

• witches, God, Nature … when involved in an explanation, these entities are frequently treated as ontologically basic mental entities, creating a false sense of simplicity

Why?

• Help conservation of expected evidence. Hard to conserve when you don't see the whole tree of possibilities, only one branch as it plays out.
• Blinding (same rationale as pre-registered research)
• Assure others that you did not do Texas sharpshooter, data-dredge etc.