How Chord Loops *Really* Work

Get Nebula using my link for 40% off an annual subscription: go.nebula.tv/12tone Some additional thoughts/corrections: 1) I'm not sure I'd say most jazz songs are best understood through linear harmony: The story of the development of chord loops inevitably leads back to repeating jazz forms, after all. My point is more that jazz progressions fit that model well enough, so jazz analysts were able to retrofit those linear approaches and wind up with useful insights, even as the harmonic structures started to look a little more rotational. 2) I should note that, when I say that picking the right loop is a balancing act between stylistic appropriateness and explicit familiarity, I'm not saying the authors necessarily did that exactly correct: I think their choice was reasonable, but it has consequences, as discussed later in the video. I was just mentioning it to explain why the process isn't as simple as "Pick the most common loop." 3) Objectively speaking, 150 participants isn't actually amazing for a study, but music cognition research doesn't tend to get a lot of funding so the studies they do are chronically undersampled as a practical necessity. That's the context I'm evaluating this under: It's not that that's the ideal number of participants, it's just that, for this particular field, it's kinda the best we tend to get, and while it still needs to be taken with a grain of salt, it's enough to be worth paying attention to the results. 4) When I talk about the ratings for different chords in the loop, I'm normalizing them all to the original key of C major. They tested it in different keys as well, but for the data analysis portion they transposed all the results back to that key for consistency, so I followed suit. 5) I'm not actually entirely sure that an emphasis on the last chord would be a good representation of a linear hearing of the loop, it's just the most apparent indicator I could think of that wasn't also covered by the previous results. Because the study was designed to test the tonal and metric hypotheses specifically, I needed something that wouldn't show up in those two, and that seemed most plausible. 6) To the extent that the data does show a bias toward the final chord, that could also probably be explained by the fact that it was the most recent thing you heard, not necessarily because of its metric position. Again, without testing what happens when you stop the loop in the middle, continuity effects look a lot like metric effects. For what it's worth, the results of the second study don't seem to show a clear last-chord bias, although the structure of that experiment makes it hard to tease out secondary preferences so I don't want to read too deep into that. 7) We can also understand the question of bimodal stability without reference to a key center by looking directly at the motion of the chords: When the two roots are complementary, the motion between them is weak (or, at least, it would be if they were consecutive) and when they're contradictory, the motion is strong. This makes a fair amount of sense, in that strong motion implies directionality and directionality subverts stability. That's honestly probably a better way of conceptualizing it because it doesn't require straying outside the model, I just went with the key-based explanation because it was simpler to explain quickly. 8) I should also clarify that, when I say the loop is the best-case scenario for major-scale bias… that's kinda the point. They were testing if that bias could appear in chord loops, so giving it the best possible chance of appearing is good study design.

7:11 "the metric hypothesis. this predicts that listeners will prefer to hear the loop in Kilometers" caught me off-guard

I didnt expect to hear about a peer reviewed study about chord loops when I woke up this morning, but here we are

I think an interesting follow up would be to take the loop but instead of finishing with a random chord from the loop, finish with a chord outside of the loop’s key. See what chords that aren’t “correct” in a classical theory sense end up with a positive outcome in a scientific setting.

I wonder if there would be instrumentation bias. Chord loops are usually performed in a different style. I notice that hesring these chord in the actual songs sounded a lot better than just piano chords. The piano might bias towards a more classical interpretation.

As a data analyst with some science & stats training, I really appreciate you being clear about the scientific and statistical approach & issues, like having a clear, pre-stated hypothesis.

I taught chord loops to my sophomore class this semester at the Frost School of Music and I have to say that I hear these loops in a linear fashion with a bunch of half cadences if ending on V and plagal half if ending on IV and deceptive if ending on iii or vi. I also hear a modal half cadence if the phrase ends on flat VII. But I like your theory!

I can't not keep hearing Fine Young Cannibals - She Drives Me Crazy, just transposed down a step. I guess the point about relating the chords to a specific piece is hard to get away from.

A friend had me do a distortion study a week or so ago, so I asked him to send you his paper when he’s done with it.

I don't hear four chord loops as either rotational or linear but rather as two distinct pairs. You can hear it more easily by taking any four chord loop and adding rest bars between bars two and three and bars four and one. They all work like this in my ears, an intake of breath before the three and before the one. This explains why I have more trouble hearing repeating three or five chords progressions as loops. It doesn't explain why I have trouble hearing repeating six chord progressions as loops since I hear them as three sets of pairs as well.

I'd love to see a version of the experiment where you fade the chord loop in and let it cycle a couple times before fading back out. Then ask which chord they think is the I. Maybe take four chords and go through every way you can order them, and see if that changes anything.

Really, I think the correct answer here is that it's a blend of linear and rotational, in a sense. I think your theory of signposts and destinations is a useful theory, but I also think that when people (well, when I) listen to music with chord loops, they're actually getting linear harmonic information that they have to reinterpret it as a loop. The chords here are C - F - Am - G. We can give them Roman numerals if we take the assumption that these Roman numerals are meaningful in the first place, right? In C, they're I - IV - vi - V. I is home, IV is an exciting place away from home, vi is an alternative home, and V is a return towards home, which arrives at the I in the next iteration of the loop to give us I - IV - vi - V - I. That is a perfectly intelligible tonal phrase, analyzable using your conventional phrase structure analysis: I - IV - vi is the Tonic region, V is the Dominant region, and the concluding I is the return of the Tonic region: T - D - T. Why would you not hear that just because the phrase is getting repeated a bunch of times? I see no reason why this wouldn't be the default interpretation, assuming the loop starts on the C (which you would get from phrasing cues in the real music). What happens if you start on F (and analyze the progression in F)? You now have F - Am - G - C - F, which is I - iii - V/V - V - I. That's... pretty legit, actually, isn't it? You can still use the standard phrase analysis, with I - iii being T, V/V being PD (or S depending on how you label phrase functions), V being D, and I being the return of T. Only problem is, the V/V is a little too strong and pulls towards C. ii would be better. A decent melody can properly tonicize the F, though. So what if you start on Am (and analyze in Am)? Am - G - C - F - Am becomes i - bVII - bIII - bVI - i. That works too, but now you have i - bVII - bIII as your T, bVI as a PD, and i as a T, making a plagal cadence. This is outside the classical phrase model, but it's normal enough for popular music. Finally, doing the same with G gives G - C - F - Am - G, which becomes I - IV - bVII - ii - I. This one... eh. ii - I is one of the weakest cadences you can have (ii7 - I, maybe it could work, but not ii - I). If we analyze in C instead, we have V - I - IV - vi, no need to loop, and it makes sense, tonally. I think, based on my own experience, that your ordinary, everyday tonality (not a reference to Tagg, but anyway) usually wins when it comes to chord loops in popular music, and you don't need fancier forms of analysis to discuss the tonal center. It's just... whatever's established by the melody anyway. The chords hardly matter. But, you know, when they do, it's usually just plain tonality. The times when that stops being the case are when the tonality is left ambiguous or the music isn't actually very tonal. The thing is, the tonality being left ambiguous works in chord loops specifically because you can do this kind of thing where you tonicize different parts of the progression. This chord loop that we're looking at is a decent example, where there are three pretty good options for the tonal center, and the music could very well not pick just one. The language of signposts and destinations is a great way to talk about the... energetics? Is that the right word? ...of the music, how it moves from harmonic place to harmonic place and such. But I think that tonality is a separate concept, and people are going to try to infer tonal functions from the chords using tonal listening, taking "tonal" in the extended sense (like accepting plagal cadences and modes and such as essentially tonal). They're not going to take tonality from the taxonomy of chord motions, which are tonal-context-dependent anyway. I'm not sure there's a way to design a study to check *that*, though. I think you should keep using your chord loop theory, because it's a useful theory. It's not *wrong*. It just maybe doesn't explain what some people may think it explains.

I would be very interested to see a study that compares this 1) based on the type of music the music students are studying/writing, and 2) comparing the results of students in formal study vs experienced self-trained musicians. This is what I am thinking of, and to be honest it's part of something that is my music theory obsession the way four-chord loops are yours. When I hear "major key bias" I think something different: "Classical" and jazz are both major key based, and if music theory doesn't seem interested in using "classical" as the default of correct and intelligent music, they use jazz as "correct and intelligent" and base all the music theory off of the major and in serving the major. THe thing is, a lot of people in more modern styles are based more in minor than in major. I have written a few pseudo-"classical" things but what I have primarily written (metal, grunge, punk, goth, industrial, witch house, etc, experimental) and a lot of the other thigns which are the main output of music today (hip-hop, dance, electronic, much of pop) uses or defaults to the minor key. And here is the thing: Writing/listening way more in minor than major your way of perceiving music is different. In both my composing and my listening, I have found many different ways to resolve (i.e. release tension and reset emotion) I neither need, nor miss the major 5 chord. I don't need or miss a leading tone. I've been resolving whenever I damn well please for quite some time now . .. just in a way that makes both "classical" and jazz fans/theorists to dismiss my music as childish unintelligent noise. Let's face it, jazz has become the new "classical."

Great video. To analyse chord loops I always first take a look at the baldest loop. Using just two chords like "Waiting for the Man" or "Heroin" (Velvets) for example. I think the first chors is always a little more the home but the major scale doesn't have to be what we are looking for (bluesmusic for example). And especially in a chord loop the feeling of being home isn't necessarily relevant. ("Stay with me" by Rod Stewart)

When you played the four music examples of the 4 different starting points, I couldn't help but be reminded of modal transpositions—most prominently with "More Than A Feeling" and "Stitches", but frankly with the other two, as well. C and A felt like Ionian & Aeolian equivalents of the rotation, obviously. But that weird overbright, "majorly major" feeling that Lydian has always had for me shows up in the selected clip of "Made Alive", and the more laid-back major feeling of Mixolydian shows up in "Sooner or Later". It probably helps that the instrumentation amplified those feelings. If we're talking about the seven modes, there's a certain argument that they're really all the same key/mode. The only thing that makes them different is how you draw attention to the specific note you've chosen as the "center", but if you don't do a good job nailing that center, then someone may just as easily hear it as a different mode and not be entirely wrong. (I'm especially looking at you, Locrian!) You raised some of these points when talking early on about variables the study was trying to control for, and in the context of this sort of "early-days" research, that's probably best for sanity. But I would love to see some of the later studies that attempt to juggle several of these variables, to see what shakes out.

I think you are right to say that interval relationships surrounding a chord are the biggest influence on our perception of it, because a chord doesn't feel like much of anything outside of it's contextual relationships. However I do think what that chord feels like depends on more of a web of context than just the one chord before and one chord after it. It's likely that the effect one chord has on another decreases as they get farther apart, both in distance through time and in how many chords separate them. The idea that major is default or that there is some greater pull towards the root seems like it could easily emerge from cultural and music theory education biases in participants, which is not helpful in answering this question from a psychoacoustic perspective. Removing these variables from participants is probably impossible too. What the brain does to identify a key center in a chord loop likely comes from a complex web of contextual note relationships and an individuals lifelong developed intuitions. The glaring problem with this study is that not much is really being isolated; and in an attempt to isolate certain variables, other variables are added in to the mix. How could they possibly tell what is causing a participant's answer when there are many possible causes for that answer being tested at the same time. There are also very few variations of the tests, which makes eliminating variables even harder. It's also very likely that multiple of these many variables are contributing to peoples answers, making isolation even more difficult. Interesting jumping off point, but to get anywhere close to an answer, new studies need to be conducted with a much larger and more diverse sample size and a stronger focus on what what is actually happening in the brain.

So just before watching this, I was viewing a video explaining the 3-body problem of gravity. The three body problem looks at the fact that with only two bodies in space, a model can be built to show how they will react with each other if parameters were slightly changed, but but when adding a third body in space, it is mathematically impossible to account for all the variables when changes to parameters were slightly altered. It is not just chord progression, it is timbre, cadence, one instrument or several, which instrument(s), arpeggio or unison notes, and I am sure there is more I have not thought of yet. Not only are you not asking the same question as the author, I am not entirely sure the original question is valid. This exercise is the 3-body problem on steroids.

It's wonderful to see you reflecting on prior ideas and willing to analyze them from different angles. It's also fun to see how many new doodles you've added to your repertoire over the years ☺️

I think perhaps part of the reason the F chord might sound less stable is because in the key of C major (or A minor, for that matter, but you get what i mean) the F chord is the root of the relative lydian mode, which is known to be an inherently unstable mode to work within, and the fact that there's such a prominent G major chord, the II chord in lydian, in the progression, I think this may contribute to how unstable the F chord is. I think it might be that when you hear a major chord and then another major chord a whole step above it, even if they may be separated somewhat, you're ears are instinctively pulled towards hearing them as IV and V in the relative major, or maybe even bVI and bVII in the relative minor, but hearing it as I and II would force us into lydian, which again, is an incredibly unstable mode. I know strict tonal analysis isn't everything of course, and this probably isn't the whole reason why the F chord sounds less stable. Still, though, I thought it might be worth bringing up, let me know if this sounds plausible or not, I could be totally wrong for all i know.

I’m so glad you used “raises the question” instead of misusing “begs the question,” which has become so prevalent elsewhere. (I know this is a usage shift in progress and I should just accept it, but it still bruises my ear every time I hear it.) Also, great discussion of chord loops. I love your channel.