Why does the $1/2$ spin of fundamental fermions (electrons, quarks, and neutrinos) split them into three variants that differ only in mass, while the integer spins of massless fundamental bosons (e.g. photons and gluons) causes no such splitting?

If the bosons had mass, would they also split into three generations?

  • 1
    $\begingroup$ Some bosons do have mass, e.g., the Higgs. And there is a real possibility that the Higgs actually is a multiplet with multiple masses. A more general issue is what framework one could use to answer this type of question. You used the string-theory tag -- do you want a string-theory answer? One could also try to answer this based on the anthropic principle, in which case I suspect the answer is that there is no answer, since I doubt it matters anthropically that we have the tau lepton. $\endgroup$
    – user4552
    Oct 13, 2014 at 1:57
  • $\begingroup$ +BenCrowell, if string theory has possible insights on that one I'd be delighted to hear them! Also, your point about Higgs is fascinating — has anyone actually postulated a Higgs triplet? $\endgroup$ Oct 13, 2014 at 2:06
  • 1
    $\begingroup$ There are some good articles about the Higgs at this blog: profmattstrassler.com $\endgroup$
    – user4552
    Oct 13, 2014 at 2:08
  • 1
    $\begingroup$ Heuristically, I think of it like this: The fermions are matter particles, and must obey no particular restrictions. The fundamental bosons arise through gauge symmetries, and there's only "one" (Lie-algebra valued) gauge field for each such symmetry. Having three generations of photons would mean having three different unbroken $\mathrm{U}(1)$ symmetries, for example. $\endgroup$
    – ACuriousMind
    Oct 13, 2014 at 12:06
  • 1
    $\begingroup$ @TerryBollinger : The "simplest" example are orbifold models. They have some number of fixed points and states are localized around these points, so the discrete number of generations arises this way. Another example is free-fermionic models. Have a look at figure 1 in this technical paper: arxiv.org/abs/arXiv:1403.4107 to see that they can get thousands of models with a given number of generations within their framework. $\endgroup$
    – Heterotic
    Oct 19, 2014 at 9:41

2 Answers 2


The number of gauge bosons is restricted by symmetry: a given theory with a certain gauge invariance admits as many gauge bosons as there are generators of the corresponding gauge group. For example, there is one generator for $\mathrm{U}(1)$, resulting in the existence of a photon. $\mathrm{SU}(3)$ admits eight generators, which yield eight gluons. This is true without any reference to generations. Regarding the Higgs, it is not entirely clear that there exists only one particle.

The number of fermionic (matter) particles in the Standard Model is not dictated by gauge symmetry and in principle one could construct a model with more or fewer generations. The question as to why there are exactly three generations is still an open problem, even in string theory.

  • $\begingroup$ If supersymmetry exists, then we have a whole bunch of other bosons out there. How does that fit with this argument? Would there be no contradiction with this argument because things like selectrons and squarks aren't gauge bosons? $\endgroup$
    – user4552
    Oct 13, 2014 at 15:50
  • 1
    $\begingroup$ The symmetry argument holds true for gauge bosons. To my understanding, there would be as many generations of superpartners as there are for quarks and leptons. $\endgroup$ Oct 13, 2014 at 16:30
  • $\begingroup$ If I could just suggest: Perhaps the more interesting facet of the symmetries question might be why, if they do exist, they are so badly broken. The breaks are like windows (or perhaps detectors) into some broader reality where things aren't as smooth as simple symmetries would suggest... and that has to make them interesting and insightful, assuming we can ever figure out the root causes of the breaks. $\endgroup$ Oct 15, 2014 at 14:12

There are two things that define a particle physics model (at low energies). The first one is the gauge group G we want the model to be symmetric under. For the Standard Model (SM) we set this to $G=SU(3)\times SU(2)\times U(1)$ (for good experimental reasons!). This will uniquely determine the number of gauge bosons needed to make the model consistent.

The second ingredient is the matter content, i.e. how many fermions and scalar fields we want the model to include. In the SM we choose 3 generations of fermions (this includes the leptons, quarks and their CP partners) and 1 scalar field (the Higgs).

From a low energy point of view, the two choices are completely independent and we can choose anything we like. For example it is conceivable to have a model with $SU(5)$ gauge symmetry and no matter at all, etc...

However, the OP is not alone in not being satisfied with this apparent lack of symmetry in treating fermions and bosons. Most physicists would indeed agree that we should try and find a more unified description and hope that nature has such a bigger symmetry at a deeper level. Theories beyond the Standard Model are indeed following such an approach. The best studied example are supersymmetric theories in which every boson/fermion has a superpartner that is a fermion/boson. In this theories, the symmetry of numbers is restored!

All in all, to answer questions like the number of generations in the SM, the masses of the fermions, etc you need a candidate for a complete theory at high energies and the answers will depend on this candidate. String theory is considered to be the most successful framework for this job and string models do indeed make concrete predictions about (among other things) the number of generations we should be observing. Unfortunately, there are too many models (vaccua) to choose from and no obvious way to make the choice...

  • $\begingroup$ Your comment about string theory being capable of predicting generations, but also leading to too many models, is at the heart of why I get very frustrated with it. A formal framework is also necessarily a language in which assertions can be made, and string theory is no exception. A predictively interesting language, such as the symmetries Dirac used to predict antimatter, should only be capable of making a limited number of such assertions. The number of vacuua in string theory in contrast implies that its fundamental semantic units— string vibrations — are too rich too predict anything. $\endgroup$ Oct 14, 2014 at 11:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.