I. Introduction
A jury convicted Miguel Vaca of various violent felonies arising from a string of armed robberies. Now Vaca asks for post-conviction relief, alleging ineffective assistance of counsel during the sentencing phase of his bifurcated trial. Because Vaca’s defense counsel possessed mental health evidence detailing Vaca’s mental illness, was apprised of the jury’s interest in Vaca’s mental condition, and did not consider whether to present such records or testimony to the jury, trial counsel was constitutionally ineffective. Vaca’s motion for post-conviction relief should have been granted in part. The judgment is affirmed in part, reversed in part and remanded.
II. Facts
A. Overview
The factual basis for this Rule 29.15 motion is not disputed. Vaca committed three armed robberies in Platte County in October and November 2002. All three robberies followed the same pattern: Vaca, wearing a ski mask, rode his bicycle from his apartment to nearby locations to commit a robbery with a Taurus revolver. He robbed two businesses — Quality Cleaners and Salon North. He also attempted the robbery of a child’s birthday party in his own apartment complex, where he abducted a woman at gunpoint and discharged his revolver at a bystander. From this pattern, police were able to apprehend Vaca and find the bicycle, ski mask, and revolver in addition to other evidence linking him to the robberies.
B. Pretrial Events
When appointed counsel first visited Vaca at the Platte County jail, Dr. Anya, the county jail psychiatrist, had already been attending to Vaca. Dr. Anya suspected that Vaca suffered from mental illness and had prescribed a regimen of Traza-done, Paxil, Clonazepam, and Zyprexa. In response to both this and Vaca’s overall demeanor, defense counsel enlisted the services of Dr. Bill Geis, a licensed clinical and forensic psychologist, to determine Vaca’s competency to stand trial and whether he had diminished capacity.
Dr. Geis interviewed Vaca and administered various psychological tests. This included the Weschler Adult Intelligence Scale to determine Vaca’s cognitive ability. The results showed that Vaca had an overall IQ of 73, which was confirmed by Vaca’s school records.
Two tests designed to detect the presence of schizophrenia were also administered. Both came back suggesting a schizophrenic condition. Dr. Geis also reviewed a Social Security Administration determination assessing a disability due in part to schizophrenia, which confirmed his results. Other medical records indicated that Vaca crashed his bicycle in 1988, which resulted in head trauma and a blood clot.
*333Dr. Geis’ report also documented a violent history including incidents in which Vaca stabbed an elementary school classmate with a pencil, punched a woman who allegedly made comments about him and physically fought with family members.
Ultimately, Dr. Geis’ report concluded: The defendant has a serious mental disease — schizophrenia—that clearly could have had an impact on his ability to form rational thought and conform his behavior to the expectations of society at the time of the offense. This condition of schizophrenia is corroborated by other medical personnel and appears to have been in existence for most of his life. He also has a condition of low intelligence (borderline intellectual functioning) that could have affected his ability to understand the impact of his actions.
Neither Dr. Geis nor defense counsel remembers discussing the report with each other, but both testified that it was standard practice to discuss any prepared psychological report. Both had vague recollections of a discussion and ultimately concluded that one must have occurred.
C. The Guilt Phase
Vaca’s trial was bifurcated by the newly codified procedure for non-capital cases in sec. 557.036, RSMo Supp.2008. Although defense counsel was an experienced trial advocate, this was his first bifurcated non-capital trial.
Defense counsel’s theory during the guilt phase was complete innocence centered on misidentification by key witnesses. Part of this strategy included counsel eliciting evidence of an uncharged fourth robbery at K.C. Collectibles. Police had recovered fingerprints from that robbery location that did not match Vaca. While police initially grouped this robbery with the three charged robberies, they later excluded it as unrelated.
It is unclear from the record whether counsel intended to introduce Vaca’s mental condition to the jury during the guilt phase. However, the State succeeded in preemptively excluding any mental condition evidence before defense counsel’s case in chief, arguing that defense counsel failed to notify the State of his intention to introduce such evidence as required by chapter 552.1
The trial record does show that defense counsel tried to tease out testimony about Vaca’s mental condition during direct examination of his brother, but was met with sustained objections from the State. At no point during the guilt phase did defense counsel call Dr. Geis or attempt to submit his report into evidence.
During guilt phase deliberations, the jury sent four questions to the court:
1) Where has [Vaca] been since arrested 11-2002
2) Was [Vaca] given psychological testing
3) Had he been compliant with medications before arrest
4) Is he currently on meds
The court did not answer any of the questions. The jury returned with guilty verdicts on all the charges against Vaca.
*334 D. Sentencing Phase and Posh-Trial Motions
During the sentencing phase of the trial, the State called to the stand the robbery-victims. Each testified about how Vaca’s actions had adversely affected their lives.
In response, defense counsel called Vaca’s father. He testified that he and Vaca’s mother had only visited Vaca twice in prison. Dr. Geis was not called during this phase of the trial, nor was his report received into evidence. The jury returned with sentences of life plus 102 years, which the court ordered to run consecutively.
Defense counsel filed a timely Rule 29.11 motion for a new trial. During the subsequent hearing, defense counsel asked the court to set aside the sentences based in part on the pre-sentence investigation report, which counsel argued, “does make mention of a report by Dr. Geis which evidences historical suffering from psychological disorders.”2 The trial court did not modify the jury sentence. On appeal, the conviction and sentence were affirmed. State v. Vaca, 204 S.W.3d 754 (Mo.App.2006)
E. Post-Conviction Proceeding and Appeal
Vaca brought a Rule 29.15 motion for post-conviction relief attacking trial counsel’s effectiveness during the sentence and guilt phase. Defense counsel and Dr. Geis were the only witnesses at the proceeding.
Dr. Geis testified during the Rule 29.15 proceeding that he was available during the sentencing phase to testify and that he would have been able to expand on his report and Vaca’s history of mental illness.
Defense counsel also testified during the Rule 29.15 proceeding. Counsel stated that the State’s evidence was substantial, that Vaca’s conviction was likely, and that he knew that this case would probably be going to the sentencing phase. The following exchange between post-conviction counsel and defense counsel then occurred:
Q. Okay. Before trial did you give any consideration to calling Dr. Geis as a witness in the sentencing phase?
A. In the sentencing phase, no.
Q. Okay. Now ... did the idea to call Dr. Geis at sentencing not even occur to you or did you rule it out for some strategic reason?
A. There was no strategic reason that I can recall for not calling Dr. Geis at sentencing.
The trial court overruled Vaca’s motion for post-conviction relief. Vaca relies on two points on appeal. First, during the guilt phase, defense counsel was ineffective for eliciting evidence of the fourth uncharged robbery. Second, during the sentencing phase, defense counsel was ineffective for giving no consideration to calling Dr. Geis to testify or to entering his report into evidence.
III. Analysis
A. Standard of Review
The motion court’s findings are presumed correct. Zink v. State, 278 S.W.3d 170, 175 (Mo. banc 2009). A motion court’s judgment will be overturned only when either its findings of fact or its conclusions of law are clearly erroneous. Rule 29.15(k); Worthington v. State, 166 S.W.3d 566, 572 (Mo. banc 2005). This Court must be left with a “definite and firm impression that a mistake has been made.” Id.
*335 B.Ineffective Assistance of Counsel
“To show that his counsel was ineffective, [Vaca] must demonstrate, first, that his counsel’s representation ‘fell below an objective standard of reasonableness.’ ” Edwards v. State, 200 S.W.3d 500, 518 (Mo. banc 2006) quoting Strickland v. Washington, 466 U.S. 668, 687-88, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). Second, there must be a showing of “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different,” which is a probability “sufficient to undermine confidence in the outcome.” Id.
C.Eliciting Evidence of the Uncharged Fourth Robbery was Reasonable Trial Strategy
Vaca contends that defense counsel’s decision to enter into evidence a fourth uncharged robbery at K.C. Collectibles was unreasonable strategy that prejudiced both his guilt and sentencing phases.
“The selection of witnesses and evidence are matters of trial strategy, virtually unchallengeable in an ineffective assistance claim.” Anderson v. State, 196 S.W.3d 28, 37 (Mo. banc 2006); Williams v. State, 168 S.W.3d 433, 443 (Mo. banc 2005). “Reasonable choices of trial strategy, no matter how ill-fated they appear in hindsight, cannot serve as a basis for a claim of ineffective assistance.” State v. Johnston, 957 S.W.2d 734, 755 (Mo. banc 1997).
Defense counsel’s overall theory alleged that the State’s witnesses had identified the wrong person and that another assailant had committed all the robberies. The fourth uncharged robbery was a building block to prove this. The police had recovered fingerprints from the fourth robbery location that did not match Vaca’s and later determined that the robbery was an isolated incident.
Defense counsel’s attempted syllogism was that the police were initially correct— all four robberies were committed by the same person, and the unmatched fingerprints were physical evidence of Vaca’s innocence. In hindsight this strategy failed, but this is not the standard for evaluating defense counsel’s effectiveness.
Defense counsel was met with overwhelming evidence of Vaea’s guilt. Police recovered a revolver from Vaca’s apartment that ballistically matched the robbery weapon. Multiple victims were all able to identify Vaca by his distinctive lisp and physical description. The victims also identified Vaca’s bicycle as similar to the one ridden by their assailant. A bag of money was recovered from his apartment. In this bag, there was a dollar bill with a brief note regarding one of the robberies in Vaca’s handwriting. Vaca initially confessed to detectives that he committed the three robberies. In the face of such compelling evidence, there were few options available for defense counsel.
As to the evidence’s effect on the sentencing phase, during the Rule 29.15 hearing, defense counsel explicitly testified that in preparing for trial he contemplated strategy’s risk. When counsel explicitly weighs the risks and rewards of admitting evidence during the guilt phase that might later be damaging during the sentencing phase, counsel’s ultimate decision is one of trial strategy. Vaca is not entitled to post-conviction relief on this point.
D.Counsel Was Ineffective as to the Treatment of Dr. Geis’ Testimony
This Court has recognized that a claim of ineffective assistance may be found when a movant can prove that “(1) trial counsel knew or should have known of *336the existence of the witness; (2) the witness could be located through reasonable investigation; (3) the witness would testify; and (4) the witness’s testimony would have produced a viable defense.” Hutchison v. State, 150 S.W.3d 292, 304 (Mo. banc 2004).
All the elements of Hutchison are met in this case. The State concedes the first three points that Dr. Geis was known by counsel, available and willing to testify. As to the fourth element, the submission of mental health history, subject to counsel’s strategic judgment, is a viable defense during a bifurcated sentencing phase. See Glass v. State, 227 S.W.3d 463 (Mo. banc 2007).
As stated earlier, “the selection of witnesses and evidence are matters of trial strategy, virtually unchallengeable in an ineffective assistance claim.” Williams, 168 S.W.3d at 443. However, this is true “[w]here counsel has investigated possible strategies.” Middleton v. State, 103 S.W.3d 726, 736 (Mo. banc 2003) (emphasis added). Here, defense counsel admitted that he gave no consideration to calling Dr. Geis and there “was no strategic reason” for not calling him.
This is not a case where counsel failed to investigate the existence of a mental illness. See e.g. State v. Ferguson, 20 S.W.3d 485, 509 (Mo. banc 2000); Williams, 168 S.W.3d at 441. Rather, defense counsel actually possessed Dr. Geis’ report, and he and Dr. Geis discussed its contents. Counsel recognized problems associated with Vaca’s mental condition the moment he met him in the Platte County jail. Counsel’s statement in the Rule 29.11 hearing shows that he knew about the report after the trial’s completion. Dr. Geis was available and willing to testify on Vaca’s behalf.
Further, defense counsel was clearly apprised that this may be in issue in the jury’s mind. The jury asked four questions to the judge during guilt deliberations regarding Vaca’s mental state. Because defense counsel conceded that conviction was probable, he knew that strategy during the sentencing phase was vital to the representation of his client.
It is unclear why counsel did not consider whether to admit Dr. Geis’ testimony. As it was counsel’s first bifurcated non-capital trial, counsel may have erroneously believed the exclusion of mental health evidence during the guilt phase carried over to the sentencing phase.3 Counsel’s inaction fell below an objective standard of reasonableness. This omission, in front of this particular jury, undermines this Court’s confidence in the sentencing phase’s outcome.
While Dr. Geis’ testimony and report may have helped Vaca, it may have also embittered the jury.4 This Court has noted that counsel strategically concluding so has been precisely reasonable in previous *337sentencing cases.5 As such, the holding of this case is not that counsel was ineffective for not calling Dr. Geis. Rather, this case rests on the fact that the question of whether to call Dr. Geis was never considered.
IV. Conclusion
Defense counsel must retain the flexibility to make strategic decisions regarding the evidence available and the circumstances of his or her client’s actual trial. Counsel may choose to call or not call almost any type of witness or to introduce or not introduce any kind of evidence for strategic considerations. Here however, experienced defense counsel candidly admitted that, without consideration and for no strategic reason, he failed to call a mental health expert. His failure to consider was ineffective.
This Court’s holding should not be construed to direct future defense counsel for Vaca to either admit or omit Dr. Geis’ testimony during his new sentencing phase. Rather, what is required is that future counsel give due consideration to the question before making a strategic choice.
Vaca’s trial counsel was constitutionally ineffective during the sentencing phase of trial. He is entitled to a new sentencing phase. The motion court’s overruling of his motion as to the sentencing phase is reversed. The judgment in all other respects is affirmed. The cause is remanded.
TEITELMAN, RUSSELL, WOLFF, BRECKENRIDGE and STITH, JJ., concur. FISCHER, J., dissents in separate opinion filed.. "Evidence of mental disease or defect excluding responsibility shall not be admissible at trial of the accused unless the accused, at the time of entering such accused’s plea to the charge, pleads not guilty by reason of mental disease or defect excluding responsibility, or unless within ten days after a plea of not guilty, or at such later date as the court may for good cause permit, the accused files a written notice of such accused's purpose to rely on such defense.” Sec. 552.030.2, RSMo Supp.2008.
. The Pre-sentence Investigation Report is not part of the record on appeal; however, mention of it is made in the transcript, which is part of the record.
. "The purpose of having a separate penalty phase in non-capital trials ... is to permit a broader range of evidence relevant to the appropriate punishment to be imposed.” State v. Prosser, 186 S.W.3d 330, 333 (Mo.App.2005). While a broader range of evidence becomes relevant during sentencing, admissibility is still subject to the trial court’s discretion. Sec. 557.036.4, RSMo. Supp. 2008.
. Commentators have noted the double-edged nature of such evidence, finding that many jurors hold a "presumption of an absolute linkage between mental illness and dangerousness.” Michael L. Perlin, The Sanest Lives of Jurors in Death Penalty Cases: The Puzzling Role of Mitigating: Mental Disability Evidence, 8 Notre Dame J.L. Ethics & Pub. Pol’y 239, 274 (1994).
. See, e.g., Strong v. State, 263 S.W.3d 636, 653 (Mo. banc 2008) (trial counsel’s failure to call two psychological experts held effective assistance because counsel sought to show defendant as a "good man” not a mentally deranged one); Edwards v. State, 200 S.W.3d 500 (Mo. banc 2006) (counsel's failure to pursue introduction of defendant's Asperger's syndrome held effective assistance because counsel sought to show defendant as a loving family person); See also Storey v. State, 175 S.W.3d 116, 144-47 (Mo. banc 2005).