Physics, from the beginning
A ruler compares lengths. A clock compares changes. A detector records a difference. Recognition Science asks how much of physics follows from taking these simple acts seriously.
Start with a comparison ↓We’ll build the idea together. You need ratios, a little algebra, and curiosity. Each diagram gives you something to try.
A model is a simplified description with rules you can check. We’ll see what follows from a rule, why RS uses it, and what gives it a physical meaning. The questions connect the lesson; each new model states the extra ingredient it needs.
First pass · about 25–35 minutes
Follow comparisons through wave addition, then take the marked shortcut to the recap.
Deeper route · another 20–30 minutes
Explore waves on the cycle and how a filter keeps a record. Pause and return whenever you like.
Allow extra time to experiment. The linked research supplies the full arguments and their conditions.
Start with something you know
Put a pencil beside a ruler. Saying “the pencil is 15 centimetres long” means its length is 15 times the ruler’s centimetre interval. The number describes a relationship.
Change centimetres to inches and the number changes. Put two pencils beside each other and which one is twice as long stays the same. That relationship is a ratio: one length divided by the other.
A is twice as long as B. B is half as long as A. Both statements describe the same pair.
RS uses recognition for a physical comparison that makes a distinction available. Think of a thermometer responding to its surroundings. The word does not require a person to look at it.
To compare “before” with “after,” something from before has to remain available. A mark, a changed state, a memory: a record lets one event matter to another. RS starts with these facts about distinctions and records, then asks which mathematical rules make them consistent.
Give a difference a size
If two lengths match, their ratio is 1. If one is twice the other, the ratio is 2 in one direction and ½ in the other. A fair measure of their imbalance should give both descriptions the same answer.
RS calls this measure the comparison cost, written J. “Cost” here is a number assigned to imbalance. It has no units; a physical model must say how it connects to an energy or a measured quantity.
x is the ratio.
Add the ratio and its reciprocal, halve the result, then subtract 1.
Twice as much and half as much both cost 0.250. Swapping the comparison leaves the cost unchanged.
Equal steps along this graph multiply the ratio by the same amount. That is why ½ and 2 sit equally far from 1.
For a ratio of 2, the arithmetic is ½ × (2 + ½) − 1 = ¼. At a match it is ½ × (1 + 1) − 1 = 0. Every other positive ratio costs more than zero.
Symmetry and a minimum alone allow many curves. RS also imposes a specific rule for combining comparisons. Ratios multiply: if A is twice B and B is three times C, then A is six times C. The cost has to obey:
J(xy) + J(x/y) = 2J(x)J(y) + 2J(x) + 2J(y).
Try x = 2 and y = 3. The combined ratios are xy = 6 and x/y = ⅔. The rule requires their costs to add to the same answer computed from the costs of 2 and 3. Here J(6) + J(⅔) = 25/12 + 1/12 = 13/6. The right side also gives 2 × ¼ × ⅔ + 2 × ¼ + 2 × ⅔ = 13/6.
This is a particular consistency condition for costs of combined comparisons. Multiplying ratios alone would not force it.
With the local scale fixed so that a small fractional mismatch has half its squared size as cost, this rule selects the displayed formula. The uniqueness result rests on composition and calibration, not just on the curve looking sensible. Read the cost paper.
From a number to a physical picture
Stretch a spring and it pulls back. Compress it and it pushes back. Near its resting length, the stored energy grows approximately as the square of the displacement. Twice the small displacement means roughly four times the energy.
The RS cost has this same quadratic shape near a match. Now imagine two fixed ends and a point between them. Compare the two positive gaps. Their mismatch is smallest at the centre.
The left gap is smaller. Moving right brings the gaps closer to a match.
A constraint is something the system must respect, such as fixed ends or a fixed total amount. Finding the lowest total cost under those constraints gives an equilibrium: an arrangement that can rest.
A bridge to physical energy
Set the fixed ends 20 cm apart. Call the left gap divided by the right gap x. For this illustrative model, choose the stored energy to be E = (1 joule) × J(x). A joule is a unit of energy. The factor of 1 joule is a physical input to this example.
Changing the energy scale changes how strongly the system pushes back. To use this for an actual object, measure its response and check whether the model describes it. The comparison rule provides a shape; physical quantities, scales, and tests give it a job in the world.
Your turn · use the model
Keep the ends 20 cm apart and move the point 5 cm right of centre. Find the left/right ratio, its comparison cost, and the energy when E = (1 joule) × J. Then change the energy scale to 2 joules: what doubles, and does the balance point move?
The gaps are 15 cm and 5 cm, so x = 15/5 = 3. The cost is ½(3 + ⅓) − 1 = ⅔. With the 1-joule scale, E = ⅔ joule.
Doubling the scale gives E = 2 × ⅔ = 4/3 joules at the same position. Every energy difference doubles, as does the restoring force at a given position. The lowest energy is still at equal 10 cm gaps: multiplying by a positive number leaves the balance point at the centre.
Try the slider at 75% to check the ratio and cost.
Check your understanding
What turns a dimensionless comparison cost into a testable energy model?
Choose an answer to check the bridge from mathematics to physics.
This is how the comparison idea begins doing physics. Assign a physical meaning to the quantities, account for their connections, then work out which arrangements are stable. A rule for motion is also needed to say how the system gets there. A swinging pendulum, for example, can pass through equilibrium and keep going.
The exact formula can also be written as J(x) = (x − 1)²/(2x). Set x = 1 + ε, where ε is a small fractional mismatch. Then J = ε²/[2(1 + ε)], approximately ε²/2. The resemblance to spring energy is exact to leading order. Equating this cost with a particular spring’s energy requires that spring’s physical scale.
Ask what can repeat at a new size
Balance describes an arrangement. Now ask how a structure could grow through a sequence of levels without inventing a different rule for each level.
Suppose three successive sizes follow one repeated scaling rule. Start with 1, multiply by a number r, then multiply by r again: the sizes are 1, r, and r².
In the RS construction, the first new level joins two earlier levels by adding their sizes. This construction requires the first join to register a distinction, so its comparison cost must be nonzero. Comparing two copies of the same level would have ratio 1 and cost zero. The requirement therefore selects different earlier levels. Ordinary addition can join equal sizes; this is the extra recognition requirement used here. At this first join, only levels 1 and r are available. So the new size is 1 + r.
But the repeated scaling rule calls that same new size r². Multiplying and adding must give the same answer. This is where the equation comes from.
1.300 × 1.300 = 1.690, but 1 + 1.300 = 2.300. The multiplied length is too short.
There is exactly one positive number that works:
r² = r + 1 φ = (1 + √5)/2 ≈ 1.618
The symbol φ, pronounced “fie,” names the golden ratio. The additive join of distinct levels and the common scaling rule are the ingredients of this RS construction. Together they select φ. The full argument also explains the common scaling rule through an update that preserves addition and the ordering of comparisons.
A hierarchy is simply a sequence of levels. Its sizes are 1, φ, φ², φ³, and so on. Each of these also equals the sum of the previous two. The point is the construction, rather than a claim that every object in nature has a golden shape.
Then the matching equation would be r² = 2r. For a positive r, that gives r = 2. The golden ratio belongs to the particular “join the two distinct earlier levels” construction. Changing the construction can change the answer.
Give a lasting relationship somewhere to live
A repeating size pattern is one question. A lasting record is another: what kind of relationship can survive while its parts move? We now explore a separate model of persistence. It represents a record by a relationship between closed loops; the golden-ratio calculation alone does not require that representation.
Picture two closed chain links. You can move and bend them, but you cannot separate them without cutting a link or passing one through the other. Their relationship survives changes of shape.
Imagine storing one bit of information as “linked” or “unlinked.” Stretching or bending the loops does not change the linked state. You can read the relationship without knowing the loops’ exact shapes. This is the job of the picture: a record that survives continuous distortion.
RS calls such persistence a kept relationship. In its loop-pair realization, the relevant fact is whether the loops remain linked. The dimension of the space matters.
In three dimensions, one loop can pass through another’s opening and close. Once linked, the two cannot slide apart while staying closed and without passing through each other.
The drawings are projections onto your screen. The four-dimensional drawing indicates an extra direction; a screen cannot display that direction literally.
In a plane, two disjoint loops cannot form this kind of chain link. In four or more dimensions, an extra direction allows the link to come apart. Ordinary three-dimensional space allows this particular persistent linking of two loops.
RS’s dimension argument connects a completed, persistent recognition to such a loop pair. With that physical identification, the linking theorem selects three dimensions. The picture explains the topological reason; the identification is part of the argument, not something the drawing proves.
Count the possibilities
Next we need a way to keep track of changes. Here is the guide’s discrete state model: one two-way choice for each of the three spatial directions. Picture left/right, forward/back, and down/up, labelled 0 or 1. These are three independent binary choices, like three switches. Each records which of two alternatives applies along one direction. It keeps the distinction while leaving out the exact distance.
This binary description is an ingredient of the model. Three-dimensional space alone does not make every physical system a set of three switches. Its purpose is bookkeeping: design an update that checks every combination of these distinctions. Once the choices are specified, we can count what a complete update must visit.
Each switch is either 0 or 1. There are 2 × 2 × 2 = 8 settings. Draw them at the corners of a cube: neighbouring corners differ in just one switch.
Visit every setting, flip one switch per step, and return to the start. Eight steps are enough. Fewer cannot visit all eight. This route is called a Gray cycle.
Start at 000.
The highlighted corner is the current setting. Follow the route and watch one digit change at a time.
This cube is a map of possible settings. It is not a drawing of a tiny material cube.
RS uses this eight-state structure in its recognition cycle. A tick means one update. Counting the states explains the number eight for this complete one-switch schedule; it does not, by itself, tell us how many seconds a tick lasts.
000 → 001 → 011 → 010 → 110 → 111 → 101 → 100 → 000
Compare each pair of neighbours, including the last step home. Exactly one digit changes. The requirement to visit every setting is what makes eight the minimum for this schedule.
Sixteen. The fourth switch can be 0 or 1 for each of the existing eight settings, so it doubles the count. The rule is 2 × 2 × …, once for each independent switch.
Give a repeating pattern a phase
Two people pushing a swing together reinforce each other. If one pushes forward while the other pushes back, they can cancel. Both the size of a push and its timing matter.
A wave has a size, called its amplitude, and a position in its repeating cycle, called its phase. Represent both with an arrow: length gives amplitude, angle gives phase. For two sine waves with the same frequency, in a model where disturbances add linearly, their arrows add too. “Same frequency” means the same number of cycles per second. “Linearly” means that the combined disturbance is the sum of the two separate disturbances.
The two waves line up. Their amplitudes add: 1 + 1 = 2.
A good place to pause
Comparisons describe differences. Constraints give a balance point. Repeating rules build patterns. Waves combine according to size and timing.
Finish the first pass with the recap →
Ready for more? The deeper route below takes about 20–30 minutes, plus time to try the controls. It asks how a whole wave pattern can move and what a record keeps. This is a new layer; you can save it for another sitting.
Now return to the eight-step cycle. Its eight settings give us eight places to keep numbers, like eight seats around a table. We add a new ingredient: an amplitude arrow at each place. The arrows describe a wave pattern across the cycle; they are different from the 0-or-1 labels that name the settings.
One tick moves every arrow to the next place, with the last wrapping to the first. For a general pattern, you must follow all eight arrows. For a specially organised pattern, the shift has a simpler description: every arrow turns by the same angle. Such a pattern is a mode, like one pure tone in a musical chord.
Places 0–7 follow the cube’s route. Each arrow has length 1. The grey dashed arrow shows its starting direction; the solid arrow shows it now.
Trace one arrow: place 0 hands its arrow to place 1. Every place does the same; place 7 wraps around to place 0. The two boxes show the same arrow before and after that move.
At tick 0, mode 1’s arrows make one full turn across the eight places. Shift once: each place receives the arrow from its previous neighbour.
Try mode 1, then mode 3. Mode 1 makes one turn around the eight places; mode 3 makes three. A shift rotates every arrow by −45° in mode 1, but by −135° in mode 3. A minus sign means clockwise here.
What could a filter keep from a pattern? Compare each place with the place halfway around the cycle. Just as two thermometers can share a background temperature but differ in their readings, these two places have a shared part and a difference part.
The guide’s odd-mode filter keeps that difference part. Subtract the opposite place’s arrow from this place’s arrow, then halve the result. Swapping the two places reverses the difference. So the retained pattern reverses after four shifts. The modes with exactly this property are 1, 3, 5, and 7.
That gives the selection a concrete job: remove the part shared by opposite halves and keep their contrast. Whether a physical detector makes this particular selection requires a model of that detector.
Mode k winds k times around the cycle. A one-position shift turns its phase by −45° × k. Four shifts turn it by −180° × k. Odd modes point backwards halfway through; even modes point the same way.
Mode 1 reverses its arrow after four ticks and returns after eight.
For each odd mode, two shifts turn every arrow by a quarter turn: clockwise for modes 1 and 5, counterclockwise for modes 3 and 7, after ignoring complete turns. Apply those two shifts again and every arrow reverses. The same reversal happens to any mixture of odd modes. This quarter-turn rule respects addition of patterns and, applied twice, gives a sign reversal. Mathematicians call an operation with these properties a complex structure: it plays the role of multiplying by i, whose square is −1. Identifying this internal sign reversal with the spin of a physical particle takes an additional physical argument.
Each shift turns every arrow clockwise by 45°. Four shifts make 180°, so every arrow points opposite its starting direction. Four more shifts complete 360°, bringing every arrow home. Try this with the step button above.
Quantum physics uses these same amplitude arrows to calculate interference. A complex number is a compact way to write an arrow with two components. The symbol i acts as a quarter turn. Two quarter turns reverse the arrow, which is the meaning of i² = −1.
The Schrödinger equation describes how quantum amplitudes change with time. For these simple modes, that change is a steady turning of phase. Choose a tick duration and assign each mode an energy that produces its phase turn. Then this discrete shift can be represented exactly by steps of Schrödinger evolution. There are multiple ways to fill in the motion between ticks: an arrow could also make extra complete turns. Matching the tick-to-tick changes alone does not select a unique continuous motion or identify a physical particle.
Distinguish motion from keeping a result
Rotating the eight-position pattern is reversible: turn it back and you recover the starting pattern. Keeping only part of it is different. Once discarded information is unavailable, you cannot reconstruct the whole original from what remains.
A setting has a binary label. A wave assigns an amplitude arrow to every setting. A mode is a special wave pattern that turns uniformly when shifted. Next we will add two modes and choose what to keep.
Imagine the setting labelled 011. Its arrow changes from pointing right to pointing up. Has its binary label changed? Have you described a complete mode?
No to both. The setting is a named place for a value. Its arrow is the amplitude and phase stored there. A mode describes a coordinated pattern of arrows at all eight places, not one arrow by itself. When a mode shifts, every arrow turns by the same amount.
If that distinction feels slippery, return to the eight-place experiment and trace one arrow before continuing.
Just as two waves add, two modes can contribute to one pattern. Add their arrows separately at each of the eight places. A component means one of those contributions, like one tone within a chord. For this eight-place model, any amplitude pattern can be built by adding its eight modes with suitable sizes and phases.
Try a simple mixture: mode 1 plus mode 0. Mode 1’s arrows turn as you go around the places. Mode 0’s arrows all point right. At place 0 they initially reinforce; at place 4 they initially cancel. Keeping only mode 1 means removing mode 0’s contribution at every place, leaving a pure mode-1 pattern.
Initially the combined arrows at places 0 and 4 are 2 and 0 along the horizontal direction. Their half-difference is (2 − 0)/2 = 1 at place 0. Reverse the comparison and it is (0 − 2)/2 = −1 at place 4. This extracts the opposing mode-1 arrows while removing the shared mode-0 contribution.
The guide’s recognition operator is a rule with these two actions: advance the pattern, then retain the odd modes introduced above. The hard-selection version sets the other modes to zero. A gentler version shrinks their amplitudes on each tick.
Shrinking and discarding are different. Multiplying by 0.7 can be undone by dividing by 0.7 if you know the exact value. Multiplying by zero cannot: every starting value becomes zero. Finite fading by a nonzero known fraction is reversible in this ideal calculation. Real loss of accessible information needs something further, such as information leaving for the surroundings or becoming too small to resolve.
Shift the pattern one place.
Keep the selected modes.
Other modes fade each tick.
Each small picture adds the two arrows at that place. Numbers show the length of their sum. The bars below show the sizes of the separate contributions. Advance a tick to shift the pattern and shrink mode 0; set the retained fraction to 0 to remove mode 0 at the first tick.
Tick 0. Both component sizes start at 1. Each tick keeps the selected component and multiplies the other by 0.70.
Bars show component amplitudes, not probabilities. The arrows above show the phases too. Very small nonzero values use scientific notation: e−4 means × 10⁻⁴. A vanishingly short bar can be too small to see while its value is still nonzero. The fade fraction is a model setting, not a measured constant.
Predict before you try
If the retained fraction is 1, what does advancing a tick do to our mixed pattern?
Choose an answer, then set the fraction to 1 and advance a tick.
In the physical picture, a measurement couples a system to something that can hold a record. The hard-selection operator illustrates the difference between reversible phase motion and discarding a component. The gradual-fading version illustrates attenuation, which means a decrease in amplitude. Fading a component alone does not specify which detector result occurs. A full measurement model must include the detector, the record, and the probability rule.
Set the fraction to 0, then advance one tick. The mode-0 arrows disappear. The sum becomes just the shifted mode-1 pattern, with length 1 at every place. The filter removes a contribution to the whole pattern, rather than deleting particular places. Reset to see the mixture again.
One quarter: 1 × ½ × ½ = ¼. In exact arithmetic, multiplying that by 2 twice recovers 1. With a fade fraction of zero, the first tick erases the component and no division can recover its starting value.
Return to the world around you
The core picture brings together ratios, imbalance, constraints, repeating structure, and wave addition. The worked energy example shows how a dimensionless comparison becomes a physical model. The deeper route adds one way to move a wave pattern and select what remains; you can return to it here.
The same questions help connect several familiar subjects:
A displaced system may have a higher energy than a nearby arrangement. How sharply that energy changes determines its restoring force. RS builds physical energy models from the comparison cost and the system’s constraints.
A standing wave on a string keeps a pattern while the string moves. This gives intuition for a persistent dynamical structure. RS studies patterns that persist because their parts affect one another in a stable way. To identify one as a particular particle, the model must also account for that particle’s measured properties, such as its mass and electric charge.
A field assigns a value to each place, like a temperature map. In the gravitational model, the value describes gravitational potential: potential energy per unit mass. RS builds an energy from comparisons between neighbouring values, using positive ratios constructed from their differences, and includes where matter is located. For small differences, its balance equation takes the familiar Newtonian form. The strength of gravity must also be supplied. The linked research works through that physical model.
A clock compares a changing system with a repeating reference. The cycle gives an ordered sequence of updates; a physical clock gives their duration. Wave phases then connect those updates with quantum evolution.
Calculated in the diagrams: reciprocal costs, a balance point, the golden-ratio matching equation, eight switch settings, phase rotation, and selection from a mixed wave pattern.
Ingredients of the RS constructions: the cost-composition rule, an additive join of distinct levels, a loop pair representing a persistent recognition, binary states, and amplitude arrows. We stated these ingredients where they entered and checked what follows. The linked research gives the fuller arguments for those constructions.
Physical predictions: identifying these quantities with an actual energy, particle, clock, or detector requires a physical model and comparison with measurements. The spring example shows that step explicitly; the research papers develop the broader connections.
The unifying move is to ask the same questions each time: What differs? What is being compared? What must stay consistent? What can change, and what record remains?
That is the intuition to leave with. Physics becomes easier to follow when a formula has a job and each new structure answers a question you already understand.
Try explaining it yourself
Choose an answer, then try explaining why in your own words.
Mathematical results: the comparison cost is selected by specified consistency and calibration conditions; the additive, uniformly scaling hierarchy yields φ; the loop-pair realization selects three dimensions; three binary choices have eight settings; the illustrated cycle and wave algebra can be checked directly.
Physical content: a theory also identifies what carries those quantities, how it moves, what a clock measures, and what a detector reads. RS supplies such constructions in its research papers. Their conclusions apply under the stated physical identifications and conditions.
The scope of this lesson: the diagrams teach these ideas and compute the displayed examples. They are not a simulation of every law of nature. The guide’s finite operator model, the core uniqueness theorems, and a complete prediction for a specific experiment are distinct claims.