Why I’m focusing on invertebrate sentience

I wrote this post be­cause I’m look­ing to get feed­back on my plans, to de­scribe a ne­glected re­search path to oth­ers, and to seek fund­ing for con­tin­u­ing to re­search in this area. I don’t have a strong aca­demic back­ground in this area (I just have a philos­o­phy BA) but I’ve been es­pe­cially in­ter­ested in the ques­tion for around six years now. I’ve now finished a con­tract with Re­think Pri­ori­ties on in­ver­te­brate sen­tience. I’ve also spent around the last nine months study­ing it. If you’re in­ter­ested in fund­ing me or col­lab­o­rat­ing please get in touch! Use the term “sen­tience” syn­ony­mously with “phe­nom­e­nal con­scious­ness.”

Do­ing re­search on in­ver­te­brate sen­tience would also tend to in­volve some re­search on ver­te­brate sen­tience be­cause much of the liter­a­ture of which crite­ria to use to as­sess sen­tience is done in the con­text of ver­te­brates. This would mean that in­ver­te­brate sen­tience re­search would also con­tribute to re­search on ver­te­brate sen­tience, though this would not be the pri­or­ity.

I’m gen­er­ally not talk­ing about cephalo­pod sen­tience be­cause cephalopods have a much stronger case for their sen­tience already, they are much less nu­mer­ous than smaller in­ver­te­brates, and more peo­ple are already con­cerned for their welfare. My po­si­tion on sen­tience is rel­a­tively similar to Luke Muehlhauser’s, as pre­sented in his Re­port on Con­scious­ness and Mo­ral Pa­tient­hood. My ap­proach is in­spired by him.

In ad­di­tion to the agent-neu­tral rea­sons I will pre­sent in this post about why we might want to fo­cus more re­sources on in­ver­te­brate sen­tience, I think I have a strong com­par­a­tive ad­van­tage in do­ing in­ver­te­brate sen­tience re­search com­pared to other things that we have strong rea­sons to fo­cus on, like AI safety re­search.


The two main ways that I see this re­search hav­ing an im­pact is by bet­ter in­form­ing us to make de­ci­sions about im­prov­ing well-be­ing and by func­tion­ing as a form of in­di­rect ad­vo­cacy. I think that it can help us bet­ter make de­ci­sions about cur­rent in­ter­ven­tions (such as hu­mane in­sec­ti­cides) as well as un­der­stand con­scious­ness bet­ter so that we can cre­ate happy minds and not cre­ate suffer­ing minds.

Though much re­search re­mains to be done on the ques­tion, hu­mane in­sec­ti­cides seem to me to have a lot of po­ten­tial as a ro­bust way of re­duc­ing suffer­ing now. The idea would be to try to get farm­ers to adopt in­sec­ti­cides that ei­ther kil­led more quickly than cur­rent in­sec­ti­cides, kil­led through a less painful mechanism, or per­haps did not kill non­tar­get species of in­sects.

Hu­mane in­sec­ti­cides as an in­ter­ven­tion seems to both have higher ex­pected value and to de­pend on fewer ecolog­i­cal as­sump­tions than other meth­ods of helping wild an­i­mals. The sign of the in­ter­ven­tion also does not de­pend on whether in­sects live net pos­i­tive or nega­tive lives. Re­search on in­ver­te­brate sen­tience is prob­a­bly com­ple­men­tary to work on get­ting hu­mane in­sec­ti­cides im­ple­mented.

One of the main ways re­search may also be able to help is as a form of in­di­rect ad­vo­cacy. I be­lieve that so­ciety should be think­ing a lot more about in­ver­te­brate sen­tience, and re­search can raise aware­ness about this is­sue. Re­search might be less di­rectly per­sua­sive then more di­rect forms of ad­vo­cacy (be­cause it is not op­ti­mized for that pur­pose), but I think that there is also less worry about back­lash from it. It also seems to me like there needs to be a com­bi­na­tion of ad­vo­cacy and re­search to sup­port that ad­vo­cacy, or else the ad­vo­cacy will not be well seen. One rea­son for this is that ad­vo­cacy is more zero-sum than re­search be­cause from the per­spec­tive of so­ciety ad­vo­cacy shifts the dis­tri­bu­tion of the pie whereas re­search in­creases it. This more co­op­er­a­tive na­ture is an­other rea­son why re­search is im­por­tant. In prac­tice is prob­a­bly a fine line be­tween pure re­search and di­rect ad­vo­cacy, but it is pos­si­ble to strike differ­ent bal­ances be­tween them.

I also be­lieve that in­ver­te­brate sen­tience re­search may be promis­ing if you ac­cept the over­whelming im­por­tance of the far fu­ture. This is be­cause in­ver­te­brate sen­tience re­search may be ap­pli­ca­ble to the ques­tion of digi­tal sen­tience. There may be many sim­ple, po­ten­tially con­scious, digi­tal minds in the fu­ture, and un­der­stand­ing if they are con­scious seems im­por­tant so that we can avoid cre­at­ing a lot of suffer­ing and cre­ate a lot of hap­piness.

When I bring this up with EAs who are fo­cused on AI safety, many of them sug­gest that we only need to get AI safety right and then the AI can solve the ques­tion of what con­scious­ness is. This seems like a plau­si­ble re­sponse to me. How­ever, there are some pos­si­ble fu­ture sce­nar­ios where this might not be true. If we have to di­rectly spec­ify our val­ues to a su­per­in­tel­li­gent AI, rather than it learn­ing the value more in­di­rectly, we might have to spec­ify a defi­ni­tion of con­scious­ness for it. It might also be good to have a failsafe mechanism that would pre­vent an AI from switch­ing off be­fore im­ple­ment­ing any sce­nario that in­volved a lot of suffer­ing, and to do this we might have to roughly un­der­stand in ad­vance which be­ings are and are not con­scious.

While some of these sce­nar­ios seem plau­si­ble to me, they are also some­what spe­cific sce­nar­ios that de­pend on cer­tain as­sump­tions about the im­por­tance of AI in the fu­ture and how the pro­cess of AI al­ign­ment might go. I think un­der­stand­ing digi­tal sen­tience may be more ro­bustly good than AI al­ign­ment re­search be­cause in a greater va­ri­ety of fu­ture wor­lds, un­der­stand­ing digi­tal sen­tience will be im­por­tant.

In my re­search I in­tend to fo­cus some­what on the ques­tion of digi­tal sen­tience, but still fo­cus mainly on in­ver­te­brate sen­tience even though I view the ques­tion of digi­tal sen­tience as more im­por­tant. This is be­cause I view in­ver­te­brate sen­tience re­search as more ro­bustly good, more tractable, less weird look­ing, and as also con­tribut­ing sig­nifi­cantly to our un­der­stand­ing of digi­tal sen­tience. If we were closer to achiev­ing digi­tal sen­tience then I would fo­cus more di­rectly on that ques­tion.

I be­lieve re­search on in­ver­te­brate sen­tience con­tributes to our un­der­stand­ing of digi­tal sen­tience and vice versa. In­deed, all re­search on sen­tience is helpful for do­ing re­search on any other kind of sen­tience. For un­der­stand­ing the pos­si­bil­ity of sen­tience in be­ings who are very differ­ent from us it is helpful to un­der­stand cases of sen­tience that are clearer to us, so that we can un­der­stand what we are look­ing for. Re­search­ing digi­tal sen­tience (and the pos­si­ble sen­tience of other en­tities such as plants or bac­te­ria) can also give us per­spec­tive that helps with our un­der­stand­ing of in­ver­te­brate sen­tience, but for now I think re­search­ing in­ver­te­brate sen­tience is more promis­ing.


In­ver­te­brate sen­tience re­search may be very ne­glected. With rare ex­cep­tions, most peo­ple do not care at all about in­ver­te­brate sen­tience. This lack of car­ing may be largely due to fac­tors that are not morally defen­si­ble, such as the fact that in­ver­te­brates look very differ­ent from us or are much smaller than us.

In­ver­te­brate sen­tience re­search is some­what less ne­glected than gen­eral efforts to help in­ver­te­brates. Some rele­vant re­search gets done out of in­tel­lec­tual cu­ri­os­ity, and quite a bit more re­search than that gets done be­cause it helps hu­mans in some way. There is a much smaller amount of re­search done di­rectly on the ques­tion of in­ver­te­brate sen­tience. I tend to think that in­ver­te­brate sen­tience re­search is more im­por­tant than cur­rent efforts to help in­ver­te­brates be­cause in­ver­te­brate sen­tience re­search may shape the far fu­ture more than in­ter­ven­tions to help in­ver­te­brates now.

In­ver­te­brate sen­tience re­search lies at the in­ter­sec­tion be­tween biol­ogy, philos­o­phy, neu­ro­science, and com­puter sci­ence to some ex­tent. This tends to mean that there are fewer ex­perts in this area than one might ex­pect. Many in­ver­te­brate biol­o­gists who might oth­er­wise have a lot to con­tribute in the area are not philo­soph­i­cally in­clined, and have not thought about the eth­i­cal im­pli­ca­tions of their knowl­edge, and so be­come con­fused about the ques­tion of in­sect sen­tience.


I don’t be­lieve that re­search on in­ver­te­brate sen­tience is very tractable. This is be­cause con­scious­ness is a thorny is­sue. Some au­thors such as Daniel Den­nett have claimed to ex­plain con­scious­ness, but they haven’t offered us plau­si­ble crite­ria for de­ter­min­ing that an en­tity is or is not con­scious. And this is what we need to find. How­ever, I do think that due to its ne­glect­ed­ness, there is some room for mak­ing more progress on the ques­tion than might be ex­pected given its in­tractabil­ity.

Luke Muehlhauser claims (and I agree) that we would be hard-pressed to as­sign a very high or very low prob­a­bil­ity to in­ver­te­brate sen­tience. This would sug­gest that there is a ceiling on how use­ful fur­ther re­search may be on this ques­tion. Muehlhauser also men­tions (and I agree) that ad­just­ing the prob­a­bil­ities you as­sign the sen­tience of differ­ent en­tities does not af­fect the ex­pected value dra­mat­i­cally. For ex­am­ple, the differ­ence be­tween as­sign­ing a 5% prob­a­bil­ity and a 50% prob­a­bil­ity is epistem­i­cally vast but ar­guably prac­ti­cally in­signifi­cant. It merely af­fects the amount of ex­pected value rep­re­sented by in­ver­te­brates by one or­der of mag­ni­tude. There are very roughly 10^18 in­sects in the world, and this num­ber is still mul­ti­ple or­ders of mag­ni­tude higher than the num­ber of ver­te­brate an­i­mals.

This would sug­gest that work­ing on in­ver­te­brate sen­tience is not im­por­tant. How­ever, there are some good ar­gu­ments that talk­ing about de­gree of sen­tience makes sense, and our best as­sess­ment of the de­gree of sen­tience of an en­tity is some­thing that more plau­si­bly can be shifted mul­ti­ple or­ders of mag­ni­tude through re­search into the ques­tion.

It also seems plau­si­ble that many of the same ar­gu­ments about why an in­ver­te­brate species might be more likely to be sen­tient also ap­ply as ar­gu­ments about why that species might have a higher de­gree of sen­tient (if they are sen­tient at all) . This is be­cause most of the best ar­gu­ments why an in­ver­te­brate species might be more likely to be sen­tient points to a similar­ity re­lated to sen­tience in us be­tween our mind and theirs. Similar­i­ties be­tween our minds and theirs are also ev­i­dence of similar de­grees of sen­tience. This means that we may not have to pri­ori­tize much be­tween these two ap­proaches. This has been my im­pres­sion so far, but I imag­ine fu­ture re­search might in­di­cate to me more ways that would provide ev­i­dence for one of these with­out pro­vid­ing ev­i­dence for the other.

Most peo­ple do not make an ex­pected value calcu­la­tion when it comes to these ques­tions, and so do­ing re­search on the like­li­hood of in­ver­te­brate sen­tience can still be use­ful to up­date and per­suade them. In my ex­pe­rience, most peo­ple typ­i­cally roughly treat the ques­tion of in­ver­te­brate sen­tience as be­ing ei­ther yes (100%) or no (0%).

What my next steps would be:

I par­tic­u­larly want to do re­search into de­gree of sen­tience and write about that. I plan to write shorter blog posts as I go be­cause it is easy enough to do, helps me con­soli­date knowl­edge, and gives me a faster feed­back loop. I will also try to work on some longer, more pol­ished, doc­u­ments. I will in all like­li­hood con­tinue to up­date the table of po­ten­tial con­scious­ness in­di­cat­ing fea­tures that will be pub­lished with the re­port on in­ver­te­brate con­scious­ness I worked on with Re­think Pri­ori­ties.