In Search oftheOrigins ofConsciousness
Simona Ginsburg and Eva Jablonka: The Evolution oftheSensitive Soul:
Learning andtheOrigins ofConsciousness. MIT Press, Cambridge, MA,
2019, 646pp, ISBN: 9780262039307
Received: 30 August 2019 / Accepted: 4 September 2019
© The Author(s) 2019
The Evolution of the Sensitive Soul is a landmark attempt to make progress on the
problem of animal consciousness. Ginsburg and Jablonka propose a general cog-
nitive marker of the presence of consciousness: Unlimited Associative Learning.
They use this marker to defend a generous view about the distribution of conscious-
ness in the natural world, on which a capacity for conscious experience is com-
mon to all vertebrates, many arthropods and some cephalopod molluscs. They use
this inferred distribution to defend a view about the evolution of consciousness, on
which it has evolved at least three times, ﬁrst evolved at around the time of the Cam-
brian explosion (just over 500million years ago), and was in fact the driving force
behind that explosion. In this essay review, I reﬂect critically on the book’s central
idea: the proposal that Unlimited Associative Learning provides a general marker of
If I throw a ball across a ﬁeld, there are facts about the physical processes going
on as the ball ﬂies through the air, but there is nothing it feels like from the ball’s
point of view. The ball feels no joy or pain; it doesn’t experience the rush of the air
or the colour of the sky. Now imagine a dog chasing the ball. There will again be
facts about the physical processes going on as it sprints across the grass. But this
time, there will also be something it feels like from the dog’s point of view. To be
conscious in the most basic sense, or sentient, is to be like the dog rather than the
ball. It is to be the kind of thing that has experiences, be they experiences of colour,
shape, odour, pleasure, pain, joy or frustration. Somewhere in nature there is a line
* Jonathan Birch
1 Department ofPhilosophy, Logic andScientiﬁc Method, London School ofEconomics
andPolitical Science, London, UK
between the entities that have no experiences of any kind, and those entities that do
have some experiences of some kind. Finding that line, and understanding how it
was crossed, is a challenge for evolutionary biology.
It’s tempting to ask: how can we possibly understand the evolution of conscious-
ness if we don’t understand the nature of consciousness? And isn’t understanding
the nature of consciousness a notoriously hard problem? So hard it has come to be
known as “the hard problem”? This didn’t deter many of the luminaries of nine-
teenth century evolutionary thinking, including Jean-Baptiste Lamarck, Herbert
Spencer and George Romanes. But, with a few exceptions, it did deter most of the
luminaries of twentieth century evolutionary thinking. For a long time, the received
wisdom has been that evolutionary biology should remain silent on these matters,
leaving them in the capable hands of philosophers and other full-time speculators.
Could it be that the time for silence has now passed? The last 30years have seen
the rise of a science of consciousness, aimed at uncovering the neural mechanisms
involved in conscious experience, drawing on the technological advances of mod-
ern neuroscience. Opinions diﬀer on how much progress the neuroscience of con-
sciousness has made. The science is still often described as being in an immature
or “maturing” phase (Wiese 2018; Michel etal. 2018). We don’t have a universally
agreed set of signatures of consciousness or a standardised method for detecting
them. Debate still rages as to how to look for consciousness in subjects who cannot
verbally report their experiences, such as non-human animals and patients in a mini-
mally conscious state. Yet it is widely agreed that the ﬁeld has made at least some
What sort of progress? Various potential neural signatures (or neural correlates)
of consciousness have been identiﬁed (reviewed in Dehaene 2014, Chapter4). A
speciﬁc type of neural oscillation (the gamma wave) has long been thought to be
important, although not everyone agrees about this. A speciﬁc type of event-related
potential (the P3 wave) also seems to matter. Conscious experience seems to result
not from localised brain activity in a speciﬁc region, but rather from brain activity
that implicates many diﬀerent regions of the cortex, as well as the thalamus. Some
take the prefrontal cortex at the front of the brain to play a special role (Dehaene
2014), whereas others emphasize a “posterior hot zone” at the back of the brain
(Boly etal. 2017).
The upshot is increasing optimism about the idea that science is ready to confront
the puzzles of animal consciousness in a serious way, and to start making evidence-
based assessments about which non-human animals are conscious and which are
not (Boly etal. 2013; Le Neindre et al. 2017). This seems clearly correct in rela-
tion to non-human mammals. Our fellow mammals have a cortex like ours (albeit
smaller), and the (thalamo)cortical mechanisms supporting consciousness in their
brains are likely to be close homologues of those supporting consciousness in ours.
It is a straightforward question whether those mechanisms are there or not, and the
evidence so far strongly suggests that they are (Boly etal. 2013; Le Neindre etal.
Yet this strategy runs into a methodological problem as soon as we look beyond
mammals, because non-mammals have no cortex. There are potentially homologous
structures, such as the dorsal pallium in birds, but even in birds the diﬀerences are
In Search oftheOrigins ofConsciousness
substantial (Güntürkün and Bugnyar 2016). The diﬀerences are greater still in rep-
tiles, ﬁsh, and invertebrates. Does this mean non-mammals cannot be conscious?
Of course not. It just means that, if they are conscious, the neural mechanisms sup-
porting consciousness are likely to be quite diﬀerent from those which support con-
sciousness in human brains. But if non-mammals can’t report their experiences, and
we can’t look for characteristic patterns of cortical activity, what exactly are we sup-
posed to look for? What are the distinctive markers of consciousness in non-mam-
mals? How can a science of non-mammalian consciousness get oﬀ the ground?
The Evolution of the Sensitive Soul is a landmark attempt to make progress on this
problem. Ginsburg and Jablonka propose a general cognitive marker of the presence
of consciousness: Unlimited Associative Learning. They use this marker to defend a
generous view about the distribution of consciousness in the natural world, on which
a capacity for conscious experience is common to “all vertebrates, many arthropods
and some cephalopod molluscs” (p. 392). And they use this inferred distribution to
defend a view about the evolution of consciousness, on which it has evolved at least
three times in these three lineages, ﬁrst evolved at around the time of the Cambrian
explosion (just over 500million years ago), and was in fact the driving force behind
The curious title is a reference to Aristotle’s distinction between nutritive, sensi-
tive (i.e. sentient) and rational souls, and this gives an initial ﬂavour of the book as
whole. Although packed with scientiﬁc detail, it also abounds with historical notes,
ﬁne quotations and illuminating digressions. The authors have a keen historical sen-
sibility and a vast range of inﬂuences. Lamarck, Spencer, Romanes and William
James feature particularly heavily. The result is a book that makes a hefty contribu-
tion to the literature, literally as well as metaphorically. The main text is 482 pages
long, with 62 pages of notes and a 72-page bibliography with over 900 entries. The
word “ambitious” does not do it justice. It would be a foolish enterprise for me to
attempt to summarise it. What I will do instead is reﬂect critically on the book’s
most important idea: the proposal that Unlimited Associative Learning provides a
general marker of consciousness.
1 Unlimited Associative Learning (UAL) asaMarker ofConsciousness
At the heart of Ginsburg and Jablonka’s approach is the idea that, to study con-
sciousness scientiﬁcally in non-mammals, we need cognitive markers of conscious-
ness. If we look at behaviour alone, assuming that behaviour resembling human
behaviour will have similar causes, we will be led down the path of unrigorous
anthropomorphism. If we look at neuroanatomy alone, giving undue weight to the
presence of a cortex, we will be led to an unjustiﬁed scepticism about consciousness
in non-mammals. We need a middle path between credulity and undue scepticism,
and the middle path is to look for cognitive markers.
Ginsburg and Jablonka argue, very plausibly, that we should look in particular
at learning. When we ask what sorts of cognitive processing might be facilitated
or enabled by conscious experience, learning is an obvious candidate, and the idea
of a link between learning and consciousness has a long history (the book includes
a quotation from Romanes positing such a link in 1883). Ginsburg and Jablonka
are well aware, however, that not any kind of learning will do as a reliable marker
of consciousness. There is evidence that a surprising amount of learning can occur
even when the stimuli are not consciously perceived.
How sophisticated can unconscious learning be? One line of evidence concerns
subliminal fear-conditioning in humans (Raio etal. 2012; Lipp et al. 2014). For
example, Lipp etal. (2014) found that subjects can learn an association between a
subliminally presented image (e.g. a picture of a wallaby) and an electric shock that
occurs on stimulus oﬀset, eventually showing conditioned electrodermal responses
indicative of fear as soon as the image appears.1 Two more recent studies have found
evidence of operant conditioning on subliminal stimuli (Pessiglione etal. 2008; Atas
etal. 2014). Severely invasive experiments on rats by Grau etal. (reviewed in Allen
etal. 2013) showed that spinally transected rats can do forms of conditioning and
avoidance learning on electric shocks to the legs, below the point of transection.
Information about these stimuli could not possibly reach the rat’s brain.
Reading all this, one starts to wonder: what’s left for consciousness to do? What
more sophisticated forms of learning are there that might still be linked to conscious-
ness? Here Ginsburg and Jablonka enter the fray with an interesting proposal. Their
claim is that conscious experience is required for Unlimited Associative Learning
(UAL), a form of associative learning with three distinctive features (pp. 230–232):
1. Compound stimuli: the conditioned stimulus can be a compound of discrimina-
ble perceptual features (e.g. a black-and-yellow buzzing object with a particular
odour). These features may be in diﬀerent sense modalities or in a single sense
2. Novel stimuli: the conditioned stimulus can be novel to the animal, in the sense
that it is “neither reﬂex eliciting nor preassociated” with an unconditioned stimu-
lus or with past reinforcement (p. 232).
3. Second-order conditioning: there is second-order as well as ﬁrst-order condi-
tioning. A conditioned stimulus can be associated with some other conditioned
stimulus or action, and so on, building up long chains of associative links between
stimuli and actions.
Ginsburg and Jablonka’s thesis, in short, is that second-order conditioning involv-
ing novel, compound stimuli is a signature of consciousness. This kind of learning
cannot happen, they claim, if the stimuli are not consciously experienced.
1 Earlier work (Soares and Öhman 1993; Esteves etal. 1994; Öhman and Soares 1998) had suggested
these associations could only be formed when the image was “fear-relevant” (e.g. an image of a spider or
a snake), but Lipp and colleagues’ results suggest the nature of the image does not matter.
In Search oftheOrigins ofConsciousness
2 The Inconclusive State oftheEvidence
How good is the evidence for a link between consciousness and UAL? It’s fair to
say that, to date, no experiment has convincingly shown the possibility of UAL on
subliminal stimuli. The ﬁndings noted above all fall short of showing UAL with-
out consciousness. For example, in the experiments on spinally transected rats, the
conditioned stimulus (a small electric shock) was simple rather than compound, the
stimulus was reﬂex eliciting rather than novel, and the conditioning was ﬁrst-order.
The Lipp etal. (2014) study described above comes close, because at least some of
the images (e.g. the pictures of wallabies) were novel, compound visual stimuli. A
recent study by Scott etal. (2018) also comes close: it purports to show unconscious
learning of a cross-modal association between novel stimuli. The stimuli were a
spoken name and a visual depiction of a profession. These are compound stimuli,
because they consist of arrangements of perceptually discriminable elements (a
name is a compound of phonemes; a picture is a compound of shapes and colours).
At most, however, these two studies show that the ﬁrst two components of UAL can
be present without conscious awareness. They do not show any second-order condi-
tioning of subliminal, novel, compound stimuli.
The hypothesis that UAL requires conscious awareness clearly deserves care-
ful experimental scrutiny. It’s too soon, at this stage, to guess which way the
inquiry will go. One possibility is that, as soon as psychologists look seriously
for unconscious UAL, they will ﬁnd it. This would not be the end of the story: it
would refute the claim that conscious experience of stimuli is necessary for UAL,
while still leaving open the possibility that conscious experience facilitates UAL,
in the sense of boosting its reliability and/or speed or transforming its features in
some other beneﬁcial way. This would be the starting point for an investigation of
how, if at all, conscious UAL diﬀers from unconscious UAL.
Another possibility is that a learning ability even simpler than UAL will be
shown to require consciousness. For example, I don’t know of any demonstration of
the possibility of second-order conditioning of any kind on subliminal stimuli. This
raises the intriguing possibility that second-order conditioning alone might already
be a positive marker of consciousness, whether or not the stimuli are compound or
novel. Second-order conditioning (but not on novel, compound stimuli) has been
found in honey bees (Hussaini etal. 2007) and even snails (Papini 2010, p. 366).
Yet I also don’t know of any experiments that have actively looked for second-order
conditioning on subliminal stimuli in humans and failed to ﬁnd it.
In short, the link between UAL and consciousness is promising yet largely
uncharted territory. The hypothesis that UAL requires consciousness must be
considered very tentative, since it has not been directly tested. At this point, a
critic might remark: if we’re still awaiting strong evidence of a link between UAL
and consciousness in humans, isn’t it premature to start using UAL as a positive
marker of consciousness in non-human animals? And isn’t it even more prema-
ture to use the verdicts delivered by that marker as the basis for an account of the
evolution of consciousness? If we do use it in this way, aren’t we just building
conjectures on conjectures on conjectures?
I have some sympathy with this imagined critic, but I also have some sympathy
with Ginsburg and Jablonka. They are trying to motivate a research programme, not
to ﬁnish one. Their strategy is to make a bold but plausible conjecture about the
type of learning that indicates consciousness, so as to get on with the task (in Part II
of the book) of building a detailed account of the evolution of that type of learning
in the context of the Cambrian explosion. This strikes me as a potentially fruitful
approach, despite its inherently speculative character. The downside risk is not so
bad: even if the link to consciousness does not survive scrutiny, they will still have
produced a rich and interesting account of the evolution of associative learning.2
3 The Taxonomic Distribution ofUAL
The evidence regarding the distribution of UAL in the animal kingdom is also tanta-
lizingly inconclusive. As noted above, Ginsburg and Jablonka say that the evidential
picture we currently have justiﬁes the attribution of UAL, and hence consciousness,
to “all vertebrates, many arthropods and some cephalopod molluscs” (p. 392). I
worry this exaggerates the strength of the current evidence, since direct experimen-
tal tests for the presence of UAL in non-human animals have not yet been done.
To get round this problem, Ginsburg and Jablonka count evidence of “proxies
for UAL” (p. 383) as evidence of UAL, where the proxies include classical condi-
tioning on compound stimuli, operant conditioning, and several other abilities the
relation of which to UAL is unclear (numerical learning, navigation learning, and
conceptual learning). Yet this only gives us evidence of individual components of
the UAL package, not evidence of the whole package. The experiments described
above give us reason to believe that, at least in humans, some individual components
of the UAL package can occur without conscious awareness, including operant con-
ditioning and classical conditioning on compound stimuli. Given this, it is crucial to
establish the presence of the whole package.
Honey bees (Apis genus) provide an interesting case. They are among the most
intensively studied of all invertebrates, and they are good learners. Honey bees can
learn associations between novel, compound visual stimuli (Schubert et al. 2002)
and, as mentioned above, they can also do second-order conditioning on olfactory
stimuli (Hussaini etal. 2007). But can they do second-order conditioning on novel,
compound stimuli? This is what UAL requires, but no one has ever looked for it. The
evidence we do have regarding the individual components of UAL raises the prob-
ability that honey bees have the whole package, but it does not provide the kind of
strong support that would be provided by a direct test for the presence of UAL. So, it
seems premature to conclude that honey bees are probably capable of UAL, and the
same could be said, I suspect, of any other invertebrate. Of course, this is compatible
with thinking that UAL may ultimately be shown to be present in cephalopods and
2 The last chapter of the book presents an entertaining dialogue with an imagined critic, who makes a
remark along these lines (p. 452).
In Search oftheOrigins ofConsciousness
arthropods, if reliable, cheap, versatile tests for its presence can be developed. It just
underlines the need for such tests.
The Evolution of the Sensitive Soul is a book to capture the imagination of anyone
with an interest in animal consciousness, whatever their disciplinary background.
Neuroscientists, cognitive scientists, evolutionary biologists, comparative psychol-
ogists, and historians and philosophers of biology will learn a great deal from it.
This is an exciting moment for animal consciousness research, with an interdiscipli-
nary community of researchers starting to coalesce in a way reminiscent of the early
days of the science of human consciousness. This emerging ﬁeld needs foundational
work: it needs people to put forward big ideas about the markers of consciousness
and its distribution in the natural world. This book rises to the challenge, oﬀering
precisely the kind of big picture that the ﬁeld needs. The idea at the heart of that pic-
ture—that Unlimited Associative Learning is a diagnostic marker of the presence of
conscious experience, and one that is widespread in the animal kingdom—deserves
further investigation. At present, the evidence is inconclusive.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 Interna-
tional License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution,
and reproduction in any medium, provided you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license, and indicate if changes were made.
Allen C, Grau JW, Meagher MW (2013) The lower bounds of cognition: what do spinal cords reveal? In:
Bickle J (ed) The Oxford handbook of philosophy and neuroscience. Oxford University Press, New
York. https ://doi.org/10.1093/oxfor dhb/97801 95304 787.003.0006
Atas A, Faivre N, Timmermans B, Cleeremans A, Kouider S (2014) Nonconscious learning from
crowded sequences. Psychol Sci 25:113–119. https ://doi.org/10.1177/09567 97613 49959 1
Boly M, Seth AK, Wilke M, Ingmundson P, Baars B, Laureys S, Edelman DB, Tsuchiya N (2013) Con-
sciousness in humans and non-human animals: recent advances and future directions. Front Psychol
4:625. https ://doi.org/10.3389/fpsyg .2013.00625
Boly M, Massimini M, Tsuchiya N, Postle BR, Koch C, Tononi G (2017) Are the neural correlates of
consciousness in the front or in the back of the cerebral cortex? Clinical and neuroimaging evi-
dence. J Neurosci 37:9603–9613. https ://doi.org/10.1523/JNEUR OSCI.3218-16.2017
Dehaene S (2014) Consciousness and the brain: deciphering how the brain codes our thoughts. Viking
Press, New York
Esteves F, Parra C, Dimberg U, Öhman A (1994) Nonconscious associative learning: pavlovian condi-
tioning of skin conductance responses to masked fear-relevant facial stimuli. Psychophysiology
31:375–385. https ://doi.org/10.1111/j.1469-8986.1994.tb024 46.x
Güntürkün O, Bugnyar T (2016) Cognition without cortex. Trends Cogn Sci 20:291–303. https ://doi.
Hussaini SA, Komischke B, Menzel R, Lachnit H (2007) Forward and backward second-order Pavlovian
conditioning in honeybees. Learn Mem 14:678–683. https ://doi.org/10.1101/lm.47130 7
Le Neindre P etal (2017) Animal consciousness. EFSA Support Publ 14:1196E. https ://doi.org/10.2903/
Lipp OV, Kempnich C, Jee SH, Arnold DH (2014) Fear conditioning to subliminal fear relevant and non
fear relevant stimuli. PLoS ONE 9(9):e99332. https ://doi.org/10.1371/journ al.pone.00993 32
Michel M etal (2018) Opportunities and challenges for a maturing science of consciousness. Nat Hum
Öhman A, Soares JJF (1998) Emotional conditioning to masked stimuli: expectancies for aversive out-
comes following nonrecognized fear-relevant stimuli. J Exp Psychol Gen 127:69–82. https ://doi.
Papini MR (2010) Comparative psychology: evolution and development of behavior, 2nd edn. Routledge,
Pessiglione M, Petrovic P, Daunizeau J, Palminteri S, Dolan RJ, Frith CD (2008) Subliminal instrumen-
tal conditioning demonstrated in the human brain. Neuron 59(4):561–567. https ://doi.org/10.1016/j.
Raio CM, Carmel D, Carrasco M, Phelps EA (2012) Nonconscious fear is quickly acquired but swiftly
forgotten. Curr Biol 22:R477–R479. https ://doi.org/10.1016/j.cub.2012.04.023
Schubert M, Lachnit H, Francucci S, Giurfa M (2002) Nonelemental visual learning in honeybees. Anim
Behav 64:175–184. https ://doi.org/10.1006/anbe.2002.3055
Scott RB, Samaha J, Chrisley R, Dienes Z (2018) Prevailing theories of consciousness are challenged by
novel cross-modal associations acquired between subliminal stimuli. Cognition 175:169–185. https
Soares JJF, Öhman A (1993) Backward masking and skin conductance responses after conditioning to
nonfeared but fear-relevant stimuli in fearful subjects. Psychophysiology 30:460–466. https ://doi.
Wiese W (2018) Toward a mature science of consciousness. Front Psychol 9:693. https ://doi.org/10.3389/
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published
maps and institutional aﬃliations.