How good is The Humane League compared to the Against Malaria Foundation?

Thanks to the ex­perts with whom we spoke while re­search­ing this re­port!

This was writ­ten by Stephen Clare and Ai­dan Goth.

1. Summary

  • In or­der to help us pri­ori­tize Founders Pledge’s re­search and ad­vis­ing efforts across cause ar­eas, we cre­ated a rough model that tries to com­pare an an­i­mal welfare in­ter­ven­tion (The Hu­man League’s cage-free cam­paigns) to a global health in­ter­ven­tion (AMF’s bed­net dis­tri­bu­tion).

  • In as­sess­ing the im­pact of THL’s in­ter­ven­tion, we iden­ti­fied three main sources of un­cer­tainty:

    • How many an­i­mal years are af­fected by the intervention

    • How much the in­ter­ven­tion im­proves each an­i­mal’s sub­jec­tive well-being

    • How much an an­i­mal’s well-be­ing mat­ters com­pared to a hu­man’s (the moral weight)

  • The last fac­tor is re­ally tough to figure out. There are good rea­sons to think the weight might be quite high, and good rea­sons to think it might be very low. That means the range of our moral weight es­ti­mates spans mul­ti­ple or­ders of mag­ni­tude.

  • For this re­port, we made spread­sheet and Guessti­mate mod­els that com­pare The Hu­mane League to the Against Malaria Foun­da­tion for a range of differ­ent as­sump­tions about the above un­cer­tain fac­tors.

  • Im­por­tantly, we as­sumed he­do­nism (sen­tient ex­pe­rience is all that mat­ters morally), that chick­ens have moral sta­tus (their ex­pe­rience mat­ters morally), and anti-speciesism (the value of an ex­pe­rience is in­de­pen­dent of the species of an­i­mal that is ex­pe­rienc­ing it). Ac­cord­ingly, this anal­y­sis does not offer an all-things-con­sid­ered view on the rel­a­tive good­ness of THL and AMF – it as­sumes a par­tic­u­lar wor­ld­view that is rel­a­tively favourable to THL.

  • In this model, in most of the most plau­si­ble sce­nar­ios, THL ap­pears bet­ter than AMF. The differ­ence in cost-effec­tive­ness is usu­ally within 1 or 2 or­ders of mag­ni­tude. Un­der some sets of rea­son­able as­sump­tions, AMF looks bet­ter than THL. Be­cause we have so much un­cer­tainty, one could rea­son­ably be­lieve that AMF is more cost-effec­tive than THL or one could rea­son­ably be­lieve that THL is more cost-effec­tive than AMF.

  • In gen­eral, if you value hu­man well-be­ing >10,000 times more than chicken well-be­ing, AMF looks bet­ter. If you value hu­man well-be­ing <300 times more than chicken well-be­ing, THL looks bet­ter. But be­tween these moral weights the rank­ing is less clear. We think there’s a good chance (at least 50%) that the moral weight falls be­tween these bounds, where fac­tors like THL’s effec­tive­ness and the bad­ness of bat­tery cages are more im­por­tant.

  • It’s very likely that we’re miss­ing key con­sid­er­a­tions that could change our es­ti­mates by or­ders of mag­ni­tude. For ex­am­ple, we haven’t tried to ac­count for moral un­cer­tainty, in­di­rect effects of the in­ter­ven­tions or longter­mist con­sid­er­a­tions.

  • Links to our mod­els:

2. Motivation

We have to de­cide how to al­lo­cate limited re­sources be­tween the best hu­man char­i­ties and the best an­i­mal char­i­ties. We work out the trade-offs in­volved ex­plic­itly or make them im­plic­itly, but we have to some­how de­cide.

3. The­o­ret­i­cal Framework

Why this is hard

  • Our epistemic sta­tus is weak. It’s hard to know what the rele­vant vari­ables are, much less how they vary among differ­ent an­i­mals. We do not have ac­cess to the in­ter­nal ex­pe­rience of other an­i­mal species. We have to try and ap­prox­i­mate what it’s like based on ob­ser­va­tions about their be­hav­ior, re­flect­ing on our own ex­pe­rience, and con­sid­er­ing the differ­ent biolog­i­cal and philo­soph­i­cal fac­tors that seem likely to shape consciousness

  • An­i­mal-fo­cused in­ter­ven­tions are less well stud­ied than hu­man-fo­cused in­ter­ven­tions, so we know much less about how effec­tive in­ter­ven­tions to im­prove an­i­mal welfare are.

  • So there are (at least) three sources of un­cer­tainty:

    • How many an­i­mal years are af­fected by the interventions

    • How much do the in­ter­ven­tions im­prove each an­i­mal’s sub­jec­tive well-being

    • How much does an an­i­mal’s well-be­ing mat­ter com­pared to a hu­man’s?

Mo­ral weight vs. moral status

  • A moral weight sug­gests how much we should value the welfare of an an­i­mal in a given species rel­a­tive to an an­i­mal from a differ­ent species. Th­ese are usu­ally defined rel­a­tive to hu­mans, i.e. hu­mans are given a moral weight of 1.

  • Mo­ral weights are differ­ent to an an­i­mal’s moral sta­tus, i.e. whether its welfare mat­ters to us. Here we fo­cus on moral weights, which re­late more to a crea­ture’s ‘ca­pac­ity for welfare.’ For farm an­i­mals, Muehlhauser’s 2017 re­port, the most in-depth treat­ment of this ques­tion we found, as­signs chick­ens an 80% chance of moral status

Can­di­dates for key vari­ables that de­ter­mine moral weight

Some of the vari­ables we came across in re­search­ing moral weights are:

  • Clock speed of con­scious­ness (sug­gested by Muehlhauser)

    • Smaller an­i­mals have faster re­ac­tion times (e.g. imag­ine try­ing to swat a fly).

    • To the ex­tent those re­ac­tions are un­der con­scious con­trol, smaller an­i­mals would ex­pe­rience more sub­jec­tive mo­ments per unit of ob­jec­tive time

    • When mea­sur­ing well-be­ing, it seems very likely we should care about the sub­jec­tive length of ex­pe­rience more

    • This up­dates us to­wards valu­ing smaller an­i­mals more

  • Ex­pe­rience in­ten­sity (Muehlhauser)

    • More ‘in­tense’ ex­pe­riences seem like they should mat­ter more

    • Hu­man ex­pe­riences seem like they’re prob­a­bly more in­tense. But ex­pe­riences of other species may be more in­tense—for ex­am­ple, it’s not clear whether our ‘lin­guis­tic thoughts’ make our ex­pe­rience more or less intense

  • Unity of con­scious­ness (Muehlhauser)

    • Refers broadly to how var­i­ous differ­ent con­scious in­puts are com­bined into a sin­gle ex­pe­rience. E.g. con­scious states over time all be­long to some per­sis­tent “self” (sub­ject unity) and con­tents of con­scious states are unified (rep­re­sen­ta­tional unity and phe­nom­e­nal unity)

    • De­bat­able how much this in­fluences moral weight, but sub­ject and phe­nom­e­nal unity re­late to sub­jec­tive ex­pe­rience which seems rele­vant to moral weight. Hu­mans prob­a­bly have more unity of con­scious­ness than an­i­mals.

  • Brain size/​com­plex­ity (To­masik)

    • Brian To­masik is very un­cer­tain, but weakly sug­gests a view where moral weight scales non-lin­early with brain size. As a rough ap­prox­i­ma­tion he sug­gests scal­ing by N^(2/​5), where N is the num­ber of neu­rons the an­i­mals has.

    • How­ever some ev­i­dence sug­gests that the cor­re­la­tion be­tween cog­ni­tive so­phis­ti­ca­tion and neu­ron count seems weak.

4. Methodology

We made a spread­sheet model that gen­er­ates cost-effec­tive­ness es­ti­mates for many differ­ent plau­si­ble val­ues and noted where the key con­sid­er­a­tions lie. We only es­ti­mated a model to com­pare our most pop­u­lar global health in­ter­ven­tion (AMF) to our top an­i­mal welfare recom­men­da­tion (The Hu­mane League’s cam­paigns for aviaries in­stead of bat­tery cages for egg-lay­ing hens). Our model is:


num­ber of ‘hen-years’ spent in aviaries rather than bat­tery cages

benefits of mov­ing from a bat­tery cage to an aviary

moral weight of chickens

how many DALYs are averted by $1 to AMF


Effec­tive­ness of THL cam­paigns in chang­ing cor­po­rate behaviour

The FP re­port on an­i­mal welfare (pub­lished 2018) es­ti­mates that THL moves 10 hen-years from bat­tery cages to aviaries per dol­lar donated. This num­ber as­sumes there is a 60% prob­a­bil­ity that com­pa­nies hon­our their cage-free com­mit­ments and that THL’s ad­vo­cacy brought these pledges for­ward by be­tween half a year and one year on av­er­age. Sim­cikas (2019)’s es­ti­mates of cor­po­rate cam­paign effec­tive­ness are higher, though not all cor­po­rate cam­paigns re­late to bat­tery cages and aviaries speci­fi­cally. His up­per bound is that 160 hen-years are af­fected per dol­lar, with a me­dian es­ti­mate of 54. Similarly, Bol­lard (2016) sug­gests cor­po­rate cam­paigns spare at least 38 hen years from bat­tery cages per dol­lar. On the other hand, Bol­lard (2019) doc­u­ments that some com­pa­nies have de­layed or re­neged on their pledges. While he re­mains op­ti­mistic, this in­di­cates the need for on­go­ing cam­paign­ing to en­sure pledges are fulfilled which would raise the ex­pected cost.

We think the 2018 FP es­ti­mate of 10 hen-years/​$ is likely a slight un­der­es­ti­mate. Across the differ­ent tabs on the spread­sheet, we model four sce­nar­ios: 1, 10, 30 and 100 hen-years af­fected per dol­lar.

Benefits of mov­ing from bat­tery cages to aviaries

We mea­sure the benefits of mov­ing from bat­tery cages to aviaries by es­ti­mat­ing how bad bat­tery cages are and then es­ti­mat­ing how bad aviaries are as a pro­por­tion of the bad­ness of bat­tery cages.

Bat­tery cages are plau­si­bly ex­tremely bad. Hens are kept in tiny spaces, live on wire racks, are un­able to move around, and are kept from en­gag­ing in nat­u­ral be­hav­iors such as root­ing, preen­ing, and so­cial­iz­ing. Pages 20 through 23 of our re­port have more de­tail. In aviaries, birds are still kept in quite cramped con­di­tions. How­ever, birds have up to 80% more space, ac­cess to lit­ter and perches, can move around, and can en­gage in more of their preferred be­hav­iors. How­ever, there is some ev­i­dence that the rate of hen mor­tal­ity is higher in aviaries. Due to this, the FP re­port es­ti­mates there is a 5-10% chance that aviaries are worse than bat­tery cages. OpenPhil’s “cur­rent – though un­cer­tain – best guess is that even with­out ad­di­tional re­forms, the U.S. tran­si­tion to cage-free hous­ing sys­tems will on net re­duce hen suffer­ing once mor­tal­ity rates have sta­bi­lized.”

There have been a cou­ple other at­tempts at quan­tify­ing this welfare change. Char­ity En­trepreneur­ship’s weighted an­i­mal welfare in­dex gives bat­tery cages a score of −57 out of −100. As of April 2020, this is the worst score on their scale. In Com­pas­sion, by the Pound, F. Bailey Nor­wood gives caged egg-lay­ing hens a welfare score of −8 (again, the worst score on his scale) and cage-free egg-lay­ing hens a score of +2 (see Table 8.2 on p. 229).

We think that life in a bat­tery cage is very likely to have a nega­tive value—i.e. we con­cep­tu­al­ize their bad­ness as a nega­tive mul­ti­plier x of 1 unit of healthy time. Plau­si­ble es­ti­mates of x could vary across sev­eral or­ders of mag­ni­tude. Bat­tery cages could be sub­jec­tively un­pleas­ant, in which case x would be close to 0. Bat­tery cages could also be truly hor­rific, with hens spend­ing their en­tire lives in ex­treme dis­tress and pain. In that case we think bat­tery cages could be −100 or −1000, mean­ing 100 to 1000 weeks of cage-free life would be morally can­cel­led-out by 1 week or 1 year in a cage.

(The ex­is­tence of a unit of “healthy life” as an up­per bound to hu­man welfare is com­mon in analy­ses based on Qual­ity and/​or Dis­abil­ity Ad­justed Life Years. How­ever, we recog­nise that there are vary­ing lev­els of “healthy life”, some of which are bet­ter than oth­ers, and that well-be­ing might not be bounded in this way. Defin­ing a unit of pos­i­tive well-be­ing is im­por­tant to our model be­cause the nega­tive units of well-be­ing are defined in terms of trade-offs with pos­i­tive units of well-be­ing. We sug­gest in­ter­pret­ing “healthy life” for a chicken as liv­ing with all needs met, no or min­i­mal fear of pre­da­tion and dis­ease-free (e.g. per­haps the best mo­ments on a very good farm an­i­mal sanc­tu­ary) and defin­ing one unit of well-be­ing as one in­stant of healthy life, un­der­stood in this way. We’re un­cer­tain about whether this is the best way to for­mu­late our model).

Our me­dian es­ti­mates, made in­de­pen­dently, were be­tween −10 and −30, but the bounds of our spread­sheet model are −0.1 to −1000. Since bat­tery cages could plau­si­bly be ex­tremely bad (e.g. well-be­ing level of around −1000), we think that the ex­pected well-be­ing of life in a bat­tery cage is lower than the me­dian.

The spread­sheet model is not in­tended to have a prob­a­bil­is­tic in­ter­pre­ta­tion, so these bounds do not rep­re­sent a spe­cific con­fi­dence in­ter­val. In the first in­stance, the spread­sheet model shows how much bet­ter or worse donat­ing to THL is than AMF given var­i­ous as­sump­tions with­out com­mit­ting to a judge­ment about how likely it is that those as­sump­tions are true. In par­tic­u­lar, the model can help to iden­tify points at small changes in as­sump­tions change which char­ity looks more cost-effec­tive. This can some­times be suffi­cient for mak­ing de­ci­sions.

While we are con­cerned by the data show­ing in­creased mor­tal­ity rates in aviaries, we do not model the sce­nario in which aviaries are worse for hens than bat­tery cages in our non-prob­a­bil­is­tic model be­cause the cost-effec­tive­ness would, of course, be in­finitely worse than AMF. In our prob­a­bil­is­tic model, we use a prob­a­bil­ity den­sity func­tion (pdf) that puts some weight (5-10%) on the pos­si­bil­ity that aviaries are worse than bat­tery cages.

Mo­ral weight: how much does a hen’s suffer­ing mat­ter com­pared to a hu­man’s?

Th­ese es­ti­mates as­sume he­do­nism (all that mat­ters morally is con­scious ex­pe­rience of plea­sure and suffer­ing), that chick­ens have moral sta­tus (their ex­pe­rience mat­ters morally), and anti-speciesism (the moral value of an ex­pe­rience de­pends only on the qual­ity of the ex­pe­rience, not on the species of an­i­mal who ex­pe­riences it).

Not much re­search has been done on moral weights and most peo­ple are re­luc­tant to give even spec­u­la­tive es­ti­mates, so we rely pretty heav­ily on Luke Muehlhauser’s work (while rec­og­niz­ing its limi­ta­tions). Ex­treme un­cer­tainty en­tails ex­tremely wide bounds, es­pe­cially be­cause some con­sid­er­a­tions push in differ­ent di­rec­tions. Some fac­tors sug­gest an­i­mals have more weight while oth­ers sug­gest the op­po­site. This is why we con­sider what would hap­pen through­out the range of plau­si­ble es­ti­mates.

Our model has moral weights rang­ing from to (i.e. from 1 hu­man = 1 mil­lion chick­ens to 1 hu­man = 1 chicken). We think it’s very likely that if an­i­mals have moral sta­tus, their moral weight is not van­ish­ingly small (say, less 1 in 1 mil­lion).

Benefits of AMF

We use GiveWell’s up­dated es­ti­mate of AMF’s cost-effec­tive­ness as a point es­ti­mate, i.e. ~$1,700 per out­come as good as sav­ing the life of a child un­der 5. We as­sume that sav­ing a life is worth about 50 DALYs. Th­ese es­ti­mates have a mar­gin of er­ror, but since they’re un­likely to be wrong by an or­der of mag­ni­tude that shouldn’t af­fect our find­ings too much.

5. Results

In our spread­sheet model, the num­bers in the cells show how many or­ders of mag­ni­tude bet­ter THL is than AMF. So if the num­ber is black (pos­i­tive), THL is bet­ter. If the num­ber is white (nega­tive), AMF is bet­ter.

The columns are moral weight val­ues, and show how chicken ex­pe­rience is val­ued rel­a­tive to hu­man ex­pe­rience. The lower bound is , i.e. 1 mil­lion chick­ens to 1 hu­man; the up­per bound is 1.

The rows mea­sure how good mov­ing from a bat­tery cage to an aviary is for a chicken. This has two lay­ers. First, there’s a range for the bad­ness of bat­tery cages, from −0.1 (close to in­differ­ence be­tween life and death), to −1,000 (ex­treme tor­ture, 1000 days of bat­tery cage life out­weighs 1 day of healthy life). Se­cond, there’s a range for how much bet­ter aviaries are, rang­ing from 0.7 (70% as bad as bat­tery cages) to -.3 (a life worth liv­ing, 30% as good as bat­tery cages are bad).

In the differ­ent tabs, we repli­cate this spread­sheet for differ­ent es­ti­mates of THL’s effec­tive­ness.

We also trans­lated these in­puts to Guessti­mate to get an ex­pected value. We de­scribe the Guessti­mate model at the end of this sec­tion.

Where are the in­flec­tion points?

  • If you think the moral weight of chick­ens is less than 110,000, then AMF is bet­ter than THL un­less bat­tery cages are ex­tremely bad

  • If you think the moral weight of chick­ens is more than 1100, then THL is bet­ter than AMF un­less THL is very in­effec­tive and bat­tery cages aren’t that bad

  • If you think bat­tery cages are very bad (-100 or worse), then THL is bet­ter than AMF un­less chick­ens have a very low moral weight (<1/​10,000) or aviaries are as bad or worse than bat­tery cages

  • If the FP es­ti­mate of THL is a lit­tle bit pes­simistic and THL’s effec­tive­ness is closer to Saulius’ es­ti­mates, then THL is usu­ally bet­ter than AMF un­less bat­tery cages are not that bad (>-1) or chick­ens have an ex­tremely low moral weight (<1/​100,000)

  • It rarely mat­ters how much bet­ter aviaries are than bat­tery cages, as­sum­ing aviaries are at least 30% bet­ter than bat­tery cages

For each sce­nario, we have graphed the line along which AMF and THL are equally cost-effec­tive given var­i­ous as­sump­tions. Above this line, our model sug­gests a dona­tion to AMF is bet­ter; be­low the line, THL is bet­ter. The y-axis is the in­verse of the moral weight, i.e. the num­ber of chick­ens equal to one hu­man, and the x-axis shows the (nega­tive) mo­men­tary well-be­ing of life in a bat­tery cage. We plot­ted graphs sep­a­rately for differ­ent as­sump­tions about how many hen-years THL af­fects per dol­lar, with two graphs per sce­nario. One plot in­cludes val­ues of bat­tery cage bad­ness from −1 to −1000 with log­a­r­ith­mic axes and one plot that zooms in on 0 to −30 with lin­ear axes. The differ­ent lines rep­re­sent differ­ent as­sump­tions about how bad life in an aviary is com­pared to life in a bat­tery cage.

Here we in­clude the plots for the sce­nario in which 30 hen-years are af­fected per dol­lar donated to THL. We have plots for other sce­nar­ios of THL’s effec­tive­ness in this ibb album.

What do our me­dian and ex­pected value es­ti­mates sug­gest?


  • Th­ese es­ti­mates are spec­u­la­tive and not stable

  • A com­bi­na­tion of me­dian es­ti­mates is not the same as the over­all best guess, e.g. , so com­bine es­ti­mates with care

  • There are po­ten­tially im­por­tant fac­tors for which our model doesn’t ac­count (e.g. rich meat eaters, non-he­do­nis­tic con­sid­er­a­tions, vari­able moral sta­tus by species, effects of cor­po­rate cam­paigns on the num­ber of chick­ens that ex­ist)

Our low-con­fi­dence me­dian pa­ram­e­ter es­ti­mates:

  • THL af­fects 30 hen-years per $

  • Life in bat­tery cages is worth around −10 to −30 units

  • Life in aviaries is prob­a­bly a bit bet­ter than bat­tery cages, but with lots of un­cer­tainty. When we trans­late our cre­dence to Guessti­mate, the model gives an ex­pected value of aviaries be­ing ~50% as bad as bat­tery cages

  • Hu­man ex­pe­rience is about 300-500 times more valuable than chicken ex­pe­rience. Our dis­tri­bu­tion would place >10% prob­a­bil­ity on each or­der of mag­ni­tude be­tween 10x and 10,000x, with tails stretch­ing out to 1M and 110 (i.e. chicken ex­pe­rience worth more than hu­mans, due to clock speed and ex­pe­rience in­ten­sity)

A Guessti­mate model with these es­ti­mates sug­gests that:

  • In ex­pec­ta­tion, THL is >100x bet­ter than AMF

  • In the me­dian sce­nario, THL is about 2-4x more cost-effec­tive than AMF

  • A 71% chance that THL is more cost-effec­tive than AMF (calcu­la­tions here)

The ex­pected moral weight in our guessti­mate model is about 0.03 (chicken ex­pe­rience is ~30x less valuable than hu­man ex­pe­rience), which might seem very high. How­ever, note that (1) we as­sume moral sta­tus and (2) if one thinks there is some prob­a­bil­ity that the moral weight is one, then there is a lower bound to the ex­pected moral weight. If, as we do here, one as­sumes he­do­nism and that chick­ens have moral sta­tus, then we think that it is difficult to rule out the chance that hu­mans and chick­ens have equal moral weights. As a re­sult, we would ex­pect a rel­a­tively high moral weight in ex­pec­ta­tion. There may be other rea­sons for car­ing about hu­man ex­pe­rience more than chicken ex­pe­rience such that an all things con­sid­ered view would be less favourable to chick­ens. We have not taken such con­sid­er­a­tions into ac­count in this anal­y­sis.

We should note some other limi­ta­tions of the Guessti­mate model:

  • It does not seem sta­ble (e.g. the num­bers change if you re­fresh the page and get a new sam­ple)

    • The bounds and dis­tri­bu­tion you choose for the bad­ness of bat­tery cages and the moral weight have big effects

  • Spec­i­fy­ing prob­a­bil­ity dis­tri­bu­tions that ac­cu­rately rep­re­sent our cre­dences for many of these vari­ables is re­ally hard

    • Our pdf for the moral weight is nei­ther log­nor­mal nor nor­mal and we weren’t sure how best to rep­re­sent it, so we’ve had to fudge the bounds a bit to get a rea­son­able approximation

    • Our pdf for how bad aviaries are has some weight that they’re worse than bat­tery cages, but also some weight that they’re much bet­ter. We’ve fudged the bounds to get an ~8% chance aviaries are worse than bat­tery cages

  • We’re not sure if we should be calcu­lat­ing the ra­tio of the ex­pected cost-effec­tive­ness of each char­ity, or the ex­pected ra­tio of the cost-effec­tive­ness of each char­ity. But this doesn’t seem to mat­ter much

  • Nev­er­the­less, we’ve played around with the bounds and dis­tri­bu­tions for key pa­ram­e­ters and un­der most rea­son­able as­sump­tions, THL is ex­pected to be >100 times bet­ter than AMF

  • Although ac­cord­ing to our model THL is much more cost-effec­tive than AMF in ex­pec­ta­tion, the prob­a­bil­ity that THL is more cost-effec­tive than AMF is rel­a­tively low, at 0.71

6. Discussion


We think THL is more cost-effec­tive in ex­pec­ta­tion than AMF given cer­tain rea­son­able as­sump­tions, but due to high un­cer­tainty we don’t think that our mod­els offer strong ev­i­dence for this claim in gen­eral.

Given our cur­rent (lack of) un­der­stand­ing of an­i­mal sen­tience and suffer­ing, one could rea­son­ably be­lieve that AMF is more cost-effec­tive than THL or one could rea­son­ably be­lieve that THL is more cost-effec­tive than AMF, even given our THL-friendly as­sump­tions.

Would fu­ture work be use­ful?

We haven’t thought about:

  • The effect of AMF on an­i­mals—i.e. the Meat Eater Prob­lem.

  • How to in­tro­duce moral un­cer­tainty into the model. We as­sume he­do­nism, but there are other moral views on which we have some prob­a­bil­ity that would pro­duce very differ­ent es­ti­mates (e.g. deny chick­ens moral sta­tus).

It seems un­likely that we’ll learn about more vari­ables that con­tribute to moral weights with­out sci­en­tific ad­vances in our abil­ity to un­der­stand con­scious­ness. How­ever, a bet­ter un­der­stand­ing of how bad aviaries are rel­a­tive to bat­tery cages could be valuable.

De­spite the sig­nifi­cant limi­ta­tions of this work, we still think it’s im­por­tant to try and make cross-cause com­par­i­sons. We’d wel­come any feed­back on how to in­ter­pret the re­sults or im­prove our ap­proach to shed more light on this ques­tion!