Cold reading is the objection raised most often against mediumship readings – and the one explained least often. The term denotes a real, well-documented craft: you can make a stranger feel that you know a great deal about them while knowing nothing at all. This article sets out where the technique comes from, how it works, why it succeeds psychologically – and where it stops being an explanation. Both halves matter: knowing cold reading lets you spot a weak reading, and knowing it precisely tells you which findings it can not account for.
What cold reading is – and is not
British author Ian Rowland, who broke the technique down into nearly 40 individual devices in his standard work The Full Facts Book of Cold Reading, defines it about as briefly as possible: cold reading means "how to talk to someone as if you're psychic". The key word is cold: the reader has no prior knowledge of the person in front of them and gathers everything they appear to know in the moment – from probabilities, observation and the sitter's own reactions.
Two related practices are frequently lumped in with it and should be kept separate:
- Hot reading: the reader has done research beforehand – obituaries, social media, prayer cards, conversations in the waiting area. That is not skill but deception. The best-documented case is US televangelist Peter Popoff, whose "divine" knowledge of details was exposed in 1986 by James Randi and electronics expert Alexander Jason as a radio transmission: Popoff's wife read out, through an earpiece, the information visitors had written on cards themselves.
- Warm reading: pre-formulated statements that fit almost anyone yet sound individual. They are the raw material of cold reading and carry the name of the circus director P. T. Barnum – more on that below.
The toolbox
The systematic account that still sets the standard comes from psychologist Ray Hyman (University of Oregon), who worked as a palm reader himself as a student. His essay "Cold Reading: How to Convince Strangers That You Know All About Them" appeared in 1977 in The Zetetic (today's Skeptical Inquirer) and lists thirteen rules. The most important:
- Confidence before content. The delivery carries the statement, not the other way round.
- Use statistics. Whatever is already true of many people in an age group, sex or region is highly likely to be true of the individual too.
- Secure the sitter's cooperation. The reading is explicitly framed as a joint task – the sitter is to interpret, add and assign meaning.
- Fish. Statements are phrased as questions; the answer supplies the material for the next "perception".
- Observe and listen. Clothing, jewellery, speech, bodily reactions – and above all whatever the sitter volunteers.
- Flatter – and tell the sitter what they want to hear. Hyman calls this the basic rule of all.
Rowland later sorted this toolbox more finely. Four of his devices recur constantly in critiques of mediumship readings:
- Barnum statement: so general that practically everyone agrees – "sometimes you doubt decisions you defend confidently in public".
- Rainbow ruse: a trait and its opposite in one sentence – "basically rather reserved, but when the mood is right you are the centre of attention".
- Jacques statement: tailored to the stage of life – "you sometimes wonder what became of the plans you once had".
- Fuzzy fact: a blurred factual claim that can be developed either way depending on the reaction – "I see a connection to the north".
Add to this shotgunning: names, initials and causes of death thrown into a hall in rapid succession until someone responds. In large groups the chance of hitting "a Michael" or "something with the heart" is close to one.
"Statements as such have no meaning. They convey meaning only in context and only when the listener or reader can bring to bear his large store of worldly knowledge." – Ray Hyman, 1977
Why it works psychologically
At its core lies an effect demonstrated by psychologist Bertram Forer in a 1949 classic of social psychology (The Fallacy of Personal Validation, Journal of Abnormal and Social Psychology 44, pp. 118–123). Forer gave 39 students an apparently individual personality test – and then handed every one of them the same 13-sentence description, assembled from a newsstand astrology book. The students rated its accuracy at 4.26 out of 5 on average and accepted a mean of 10.23 of the 13 statements as true of themselves personally.
The effect is known today as the Forer or Barnum effect and is among the best-replicated findings in personality psychology. A review by Dickson and Kelly (Psychological Reports 57, 1985, pp. 367–382) sets out the conditions under which acceptance runs particularly high: when the description is phrased favourably, when it is announced as having been produced personally for me, and when it comes from someone perceived as competent. All three conditions are routinely met in a mediumship reading – regardless of whether anything genuine is happening there.
There is also the misjudgement of probabilities: psychologist Susan Blackmore showed in 1997 that people systematically underestimate how many others share the same seemingly personal trait or experience. What feels unique is often commonplace.
What cold reading actually achieves
The honest answer is: a great deal. A practised reader can produce a session that feels accurate, personal and moving to the sitter – with no mediumistic perception whatsoever. This is not a marginal observation by sceptics; it is the reason serious mediumship research builds such elaborate designs in the first place.
The objection is therefore justified in particular settings: stage and television formats where editing decides which statements are broadcast; readings in which the medium mostly asks rather than states; statements containing no information that could be falsified. The objection has bitten inside research as well: in 2003 Ray Hyman argued in the Skeptical Inquirer that the readings presented in the early studies of Gary Schwartz closely resembled what one would expect from competent cold readers. Part of that criticism was methodologically sound – and subsequent research took it seriously.
Where cold reading stops explaining
Cold reading is not a magic word that dissolves every finding. It is a mechanism with clear preconditions: the reader needs access to the sitter, needs feedback, and needs latitude in interpretation. Remove one of these and the explanation goes with it. The research of the past 25 years has been built along exactly that line.
1. No channel: blinding
In Julie Beischel's quintuple-blind protocol the reading takes place exclusively by telephone, the medium knows only the first name of the deceased and nothing about the living sitter – never seen, never heard, never spoken to. There is nobody whose reactions could be read. Cold reading is not refuted here; it is structurally impossible.
2. The wrong type of information
Cold reading lives on high base rates. It works for moods, character traits and common life events – and it fails on proper names, nicknames, numbers, addresses, occupational details and circumstances of death in combination. The Scottish medium Gordon Smith was studied in Glasgow over years precisely because of this type of statement. And Oliver Lazar's EREAMS study makes highly specific, independently verifiable details its explicit yardstick: what counts is not the number of hits but how guessable they are.
3. The wrong moment
The strongest case is information that nobody in the room knew at the time of the reading and that was verified only afterwards. It cannot come from observation, from feedback or from research – there was nothing to observe and nobody who could confirm. Lazar derived his basic methodological decision from this thought: if you rule out cold reading via the type of information, you no longer need blinding. That is a debatable but clean position.
The control test: decoy readings
The cold reading hypothesis is tested most directly where it is turned into a prediction. If mediumistic statements were merely Barnum sentences, strangers would relate the same transcript to their own lives just as often as the intended recipient does. That is exactly what Archie Roy and Tricia Robertson tested at the University of Glasgow: 44 recipients against 407 non-recipients, later some 300 participants and 73 readings across two and a half years. Recipients identified a significantly higher proportion of statements as applicable – the probability of chance was around 1 in 1,000,000. The same logic is used by Patrizio Tressoldi in Padua: 28 mediums, 100 readings, triple-blind, p = 0.000048. His meta-analysis of 18 experiments from 14 publications arrives at p < 10⁻⁹ – with no indication of publication bias.
| Situation | Cold reading as an explanation |
|---|---|
| Stage appearance, large hall, calls from the audience | obvious and hard to exclude |
| One-to-one reading, many questions, general statements | plausible |
| One-to-one reading with names, numbers, checkable details | weak – the base rates are missing |
| Blinded study design, medium has no contact with the sitter | structurally excluded |
| Detail verified only after the reading | not applicable |
Where the criticism still holds
The evidence is not uniform, and it would be dishonest to conceal that. The best-known controlled counter-study comes from Ciarán O'Keeffe and Richard Wiseman (Testing alleged mediumship: Methods and results, British Journal of Psychology 96, 2005, pp. 165–179). Their design consistently ruled out sensory leakage and cueing – and found no effect: accuracy rates were at chance level. Such null findings are part of the picture; they are the reason meta-analyses like Tressoldi's are needed at all, rather than playing individual studies off against each other.
A second point stands as well: in an ordinary, unblinded reading, cold reading can never be excluded in the moment. That is a property of the situation, not a verdict on the medium. Anyone assessing a reading is therefore well advised to judge not the feeling during the hour, but the information that remains afterwards.
What this means in practice
For your own reading, a short and sober practice follows from all this:
- Give little in advance. No first name of the deceased, no cause of death, no photographs, no details in the booking request.
- Watch the ratio. Is the medium stating – or asking? A high proportion of questions is no proof of cold reading, but it is a signal.
- Do not confirm immediately. A "yes, exactly!" after every statement supplies material. Better: take notes and go through them together at the end.
- Look for the checkable. Names, occupations, places, idiosyncrasies, numbers – and which of them can be verified independently.
- Record it. The transcript is considerably more sober afterwards than the memory of the session.
How a serious reading typically unfolds is described in our article on the course of a mediumship reading; what to look for when choosing, in the checklist for a reputable medium. Organisations such as the Forever Family Foundation and the certification procedures of Mark Ireland have translated exactly these requirements into testing protocols.
Conclusion
Cold reading is real, well described and in many situations the most economical explanation. Knowing it makes you rightly sceptical of general, flattering, question-driven readings. But it is an explanation with preconditions – and research has learned to remove those preconditions systematically: no visual contact, no prior knowledge, no feedback, no latitude in the evaluation. Whatever accuracy survives those conditions is precisely what cold reading no longer explains. The interpretation of that remainder is open to debate. Clearing it away with the keyword "cold reading" is no longer possible.
Sources:
• Ray Hyman, "Cold Reading": How to Convince Strangers That You Know All About Them, The Zetetic (Skeptical Inquirer) 1/2 (1977), pp. 18–37.
• Ray Hyman, How Not to Test Mediums, Skeptical Inquirer 27/1 (2003).
• Bertram R. Forer, The Fallacy of Personal Validation: A Classroom Demonstration of Gullibility, Journal of Abnormal and Social Psychology 44/1 (1949), pp. 118–123.
• D. H. Dickson & I. W. Kelly, The "Barnum Effect" in Personality Assessment: A Review of the Literature, Psychological Reports 57 (1985), pp. 367–382.
• Ian Rowland, The Full Facts Book of Cold Reading, London (several editions).
• Ciarán O'Keeffe & Richard Wiseman, Testing alleged mediumship: Methods and results, British Journal of Psychology 96/2 (2005), pp. 165–179.
• Julie Beischel et al., Anomalous Information Reception by Research Mediums Under Blinded Conditions II, EXPLORE 11/2 (2015).
• Archie Roy & Tricia Robertson, Journal of the Society for Psychical Research 65 (2001) and 2004.
• Patrizio Tressoldi et al., Anomalous information reception by mediums: A meta-analysis of the scientific evidence, EXPLORE (2020); Is There Someone in the Hereafter?, OMEGA – Journal of Death and Dying (2022).
• Society for Psychical Research, Psi Encyclopedia, entry Cold Reading, psi-encyclopedia.spr.ac.uk.
For more, see our curated knowledge collection – it links to interviews and studies on mediumship research, near-death experiences and the scientific debate.